Models ranked by quality, speed, and cost.

Difficult reasoning, code changes and document work

Pick it when

Your evaluation needs a stronger candidate for difficult reasoning, code changes or document work.

Skip it when

Routine work already passes with a cheaper model.

Inputs

TextImages

API alias

opus-5-5 opus The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Artificial Analysis measures max effort with the default fallback: 58 at $5.98 per benchmark task. High effort scores 54 at $1.82. Test the extra reasoning against your own acceptance criteria.

Reasoning and coding with effort tuned to the task

Pick it when

You want to compare the newer Sonnet with Opus on code and document work at the same effort.

Skip it when

You are choosing by token price alone: it has half Opus 5.5's token rates, but at max effort it finishes a benchmark task for only about 9% less.

Inputs

TextImages

API alias

sonnet-5-5 sonnet The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Artificial Analysis measures max with the default fallback at 56 and $5.46 per task. Xhigh scores 52 at $2.01; high scores 47 at $0.88. These are separate configurations, not a maximum bill.

Hard end-to-end work across code, browsers, and professional files

Pick it when

The job is a long, tool-driven run — a browser to operate, professional files to work through, a complicated task to finish with little guidance — and you want a premium specialist for it rather than a general flagship.

Skip it when

Drafting, summaries and routine analysis: GPT-6.1 Sol scores a point lower for $0.72 a task against $3.26, and the GPT 5.6 family covers the same work at a fraction of the rate card.

Inputs

TextImages

API alias

gpt-6-astra gpt The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

At max, Artificial Analysis measures 53 and $3.26 per task. Xhigh scores 52 at $2.31. Test whether the extra reasoning improves your accepted results; neither figure is a spending limit.

Difficult reasoning and coding, with a safeguard fallback

Pick it when

The problem is at the edge of what any model clears, the run is long, and being wrong costs more than the premium rate does.

Skip it when

Routine work, or anything a cheaper frontier row already clears — Claude Opus 5.5 scores five points higher at $5.98 a task against $7.63, Opus 5 sits two points back at $5.86, and GPT-6 Astra ties it at 53 for $3.26, under half the cost.

Inputs

TextImages

API alias

fable-5-1 fable The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

"With fallback" is Artificial Analysis's label for the API's default configuration: the small share of prompts Anthropic's safeguard classifiers flag — cybersecurity and biology, fewer than 5% of sessions by Anthropic's count — is answered by a fallback Claude model, and the score above includes those answers. The dial matters as much as the model here: at xhigh Artificial Analysis measures the same 53 for $5.98 a task, at high 51 for $3.91 — Opus 5's score for a third less — and at medium 49 for $2.98, a point below Fable 5 at about a third of its cost.

Coding and multi-step agent work at GPT-6 Sol's list rates

Pick it when

You are upgrading from GPT-6 Sol and can test the same workload at the effort you would use.

Skip it when

You need a higher score on hard agentic work: GPT-6 Astra scores a point higher at $3.26 a task against $0.72, and Claude Opus 5.5 six points higher at $5.98.

Inputs

TextImages

API alias

gpt-sol-6-1 gpt-sol The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Artificial Analysis lists this version at max effort, scoring 52 at $0.72 per task. Xhigh scores 51 at $0.39 and high 50 at $0.32. OpenAI released it in its API on September 29, 2026 as gpt-6.1-sol.

Reasoning, coding and review for existing Opus 5 workflows

Pick it when

You have hard, multi-step work validated on Opus 5 and re-testing it on Opus 5.5 is not yet worth the interruption.

Skip it when

You are starting fresh — Claude Opus 5.5 scores seven points higher at lower token rates, and at high effort it still scores 54 for $1.82 a task against this row's $5.86.

Inputs

TextImages

API alias

opus-5 Names exactly this model version and never moves.

Reasoning effort

lowmediumhighxhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

This row quotes max at 51 and $5.86 per benchmark task. Xhigh measures 50 at $4.88, and high measures 48 at $3.61. Test the lower settings on your work; these averages do not establish a maximum bill.

The previous Fable, for traffic already validated against it

Pick it when

You have production traffic validated against this version and re-testing it on 5.1 is not yet worth the interruption.

Skip it when

You are starting fresh — Fable 5.1 scores three points higher on the same list price, and at high it still clears this row's 50 by a point for $3.91 a task against $8.75. Claude Opus 5.5 scores eight points higher at $5.98 a task.

Inputs

TextImages

API alias

fable-5 Names exactly this model version and never moves.

Reasoning effort

lowmediumhighxhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

The previous Sol, for agent and coding work already validated on it

Pick it when

You have work validated on GPT-6 Sol and re-testing it on GPT-6.1 Sol is not yet worth the interruption.

Skip it when

You are starting fresh — GPT-6.1 Sol scores four points higher at the same $2 / $10 list rates, with cheaper cached input, for $0.72 a task against $1.04.

Inputs

TextImages

API alias

gpt-sol-6 Names exactly this model version and never moves.

Reasoning effort

nonelowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Artificial Analysis reports 48 at $1.04 per benchmark task at max effort, versus 47 at $1.99 for GPT-5.6 Sol and 52 at $0.72 for GPT-6.1 Sol.

Long-context agentic coding that reads text, images and video

Pick it when

Long agent and coding runs over mixed inputs, where a capable closed model at a low rate beats owning the weights and Meta is an acceptable processor of your data.

Skip it when

You need weights to host — the Muse family is closed — or your data cannot go to Meta, and the Contributor discount does not change that: it is paid for with your prompts.

Inputs

TextImagesVideo

API alias

muse-spark-1-3 muse-spark The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Worth knowing

Two rate cards. Artificial Analysis costs it at Meta's Standard rate, which is the one this row prints; the Contributor tier is far cheaper in exchange for training on your prompts and replies, and has no published cost per task. Fine for hobby projects, an easy no for anything confidential.

General reasoning and coding work with OpenAI's tooling around it

Pick it when

You have work validated on GPT 5.6 Sol and re-testing it on GPT-6.1 Sol is not yet worth the interruption.

Skip it when

You are starting fresh — GPT-6.1 Sol scores five points higher at half the token rates, for $0.72 a task against $1.99, and GPT 5.6 Terra scores five points lower for about 30% less per finished task.

Inputs

TextImages

API alias

gpt-5-6-sol Names exactly this model version and never moves.

Reasoning effort

nonelowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Long coding and knowledge-work runs with web and X tools

Pick it when

You want to test the newer Grok on multi-step coding or research workflows that need web and X search.

Skip it when

Low task cost or a context window beyond 500k tokens is the priority.

Inputs

TextImages

API alias

grok-4-7 grok The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhigh We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to xhigh. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

At xhigh, Artificial Analysis measures 46 at $3.74 per task. High also rounds to 46 at $2.73. The standard token rates match Grok 4.6, but this benchmark run costs more. This row excludes the separately priced fast variant.

Frontier-class coding and security review at a low rate

Pick it when

You want strong coding and security review at a fraction of frontier prices, rented or on your own hardware.

Skip it when

You need image input, or plain MIT terms — this one is Zhipu's own GLM-5.3 License, which allows commercial use with conditions.

Inputs

Text

API alias

glm-5-3 glm The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowhighmax We send max unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Licence

GLM-5.3 License

Served by

Fireworks AI Zhipu makes it. We buy the inference from Fireworks AI, which is the name the rate card shows.

Worth knowing

Zhipu released the weights at the end of August under its own GLM-5.3 License, which allows commercial use with conditions — read the terms before you build on it. A point ahead of Kimi K3 on the index at the same cost per task to within a cent, and it streams faster; Kimi takes images, this does not.

Reasoning with access to the live web

Pick it when

You need information from this morning in the same call as the reasoning.

Skip it when

Your agents live in a terminal, or your prompts are enormous — the window is half what most of this board offers.

Inputs

TextImages

API alias

grok-4-6 Names exactly this model version and never moves.

Reasoning effort

lowmediumhighxhigh We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to xhigh. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

This row quotes xhigh at 44 and $1.75 per benchmark task. High also rounds to 44 at $1.48. Compare the effort settings on your own work before paying for more reasoning.

Long autonomous coding runs you can self-host

Pick it when

You want one of the strongest open-weights models with image input as well as text — GLM-5.3 sits a point ahead on the index but reads text only.

Skip it when

The token rate is the point — GLM-5.3 charges well under half per token for a point more on the index, though the two finish a task for about the same.

Inputs

TextImages

API alias

kimi-k3 kimi The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Licence

Kimi K3 License

Served by

Fireworks AI Moonshot AI makes it. We buy the inference from Fireworks AI, which is the name the rate card shows.

Worth knowing

The licence is Moonshot's own, not MIT or Apache. Commercial use is allowed with conditions that start at scale, including attribution for very large products and a separate agreement for large model-as-a-service operators, so read it before you build on it.

A high score per dollar, on MIT-licensed weights

Pick it when

You want a high score per dollar on MIT weights, with the image input the GLM flagship lacks.

Skip it when

Your prompts are short and interactive — it answers at length, so a finished task costs more than its rate card suggests.

Inputs

TextImages

API alias

glm-flash-5-3 glm-5-3-flash The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowhighmax We send max unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Licence

MIT

Served by

Fireworks AI Zhipu makes it. We buy the inference from Fireworks AI, which is the name the rate card shows.

Worth knowing

Artificial Analysis measures it as very verbose, so a finished task uses more tokens, and costs more, than its rate card suggests. First natively multimodal model in the GLM-5 line — 320B total parameters, 18B active.

Balanced mid-tier for everyday work

Pick it when

You have work validated on GPT 5.6 Terra, or you want a shorter wait for the first answer than GPT-6 Sol or GPT-6.1 Sol give at max effort.

Skip it when

Score or cost decides it — GPT-6.1 Sol scores ten points higher for $0.72 a task against $1.40, at a lower output rate.

Inputs

TextImages

API alias

gpt-terra-5-6 gpt-terra The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

nonelowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Multimodal agents and long coding work

Pick it when

Mixed-media inputs — text, images, audio, video — feeding an agent or a long coding session, where its two-point lead over 3.7 Flash is worth the longer wait.

Skip it when

The problem is genuinely hard: its reasoning sits below the leaders, and Google itself says it can spend more tokens and tool calls than 3.7 Flash on difficult work. If a person is waiting, 3.7 Flash starts answering sooner and streams faster for less per task. Budget with the post-introductory price in mind.

Inputs

TextImagesAudioVideo

API alias

gemini-flash-3-8 gemini-flash The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhigh We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. Our default can differ from the benchmark in quality, cost and wait.

Cheap closed model that powers the free Meta AI

Pick it when

You want a fast, capable, every-input model and Meta is an acceptable processor of your data.

Skip it when

You need weights — the Muse family is closed — or your data cannot go to Meta.

Inputs

TextImagesAudioVideo

API alias

muse-spark-1-2 Names exactly this model version and never moves.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Worth knowing

Meta publishes a "contributor" tier at roughly a fifteenth of the standard rate in exchange for training on your prompts and replies. Fine for hobby projects, an easy no for anything confidential.

The open-weights Qwen checkpoint, for teams that host their own models

Pick it when

You need the weights, and text-only input plus a licence with conditions are acceptable.

Skip it when

You plan to resell access at scale, or your inputs are not text.

Inputs

Text

API alias

qwen-3-8-a95b qwen The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumxhigh We send xhigh unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to xhigh. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Licence

Qwen3.8-Max License

Served by

Fireworks AI Alibaba makes it. We buy the inference from Fireworks AI, which is the name the rate card shows.

Worth knowing

Artificial Analysis costs it at Alibaba's own $2 / $6 — the same rate card as the proprietary Max — but the weights are downloadable, so your actual bill is whatever your host or your hardware charges. Its Qwen3.8-Max License allows commercial use with conditions, including separate authorization for large model-as-a-service and AI work-assistant businesses, so read it before you build on it.

Low-cost text and image work on MIT-licensed weights

Pick it when

High-volume text and image work that still wants step-by-step reasoning, on weights you can take with you.

Skip it when

You need audio or video input, or the last few points of quality.

Inputs

TextImages

API alias

deepseek-v4-1-flash deepseek-v4-flash The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

nonelowhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Licence

MIT

Served by

Fireworks AI DeepSeek makes it. We buy the inference from Fireworks AI, which is the name the rate card shows.

Earlier Grok for existing integrations and cost comparisons

Pick it when

You want most of 4.6 for much less per task, or you have production traffic validated against this version.

Skip it when

You need the higher general score of a newer Grok and have verified that it improves your tasks.

Inputs

TextImages

API alias

grok-4-5 Names exactly this model version and never moves.

Reasoning effort

lowmediumhigh We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Fast multimodal work at a promotional price

Pick it when

Inputs are large or mixed-media — PDFs, images, audio, video — and you want them read quickly.

Skip it when

You need frontier-level reasoning, or weights you can host yourself.

Inputs

TextImagesAudioVideo

API alias

gemini-flash-3-7 Names exactly this model version and never moves.

Reasoning effort

lowmediumhigh We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. Our default can differ from the benchmark in quality, cost and wait.

Free general-purpose model access, within fair use

Pick it when

You want a free starting point for everyday work in MindsHub Agents or through the API. Fair use and daily rate limits apply.

Skip it when

You are benchmarking, pinning a version, or otherwise need the numbers in this row to hold still. Name a model directly instead.

Inputs

TextImages

API alias

mindshub_air A rotating alias: we move it as better options ship, so the model behind it can change without a code change on your side.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Worth knowing

The benchmark figures mirror the model this alias currently uses. Cost per task is measured at that provider’s rates, not a charge for Air within the free tier. The underlying model can change; pin a version when comparing results.

Focused text and image tasks at low token rates

Pick it when

You process many focused tasks and can test the results against clear acceptance criteria.

Skip it when

Your evaluation needs the stronger reasoning of Sol or Astra.

Inputs

TextImages

API alias

gpt-luna-6 gpt-luna The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

nonelowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Artificial Analysis reports 38 at $0.07 per benchmark task at max effort, versus 37 at $0.18 for GPT-5.6 Luna.

The previous Sonnet, for routine work already validated on it

Pick it when

You have routine drafting, summaries or small edits validated on Sonnet 5 and re-testing them on Sonnet 5.5 is not yet worth the interruption.

Skip it when

You are starting fresh — Claude Sonnet 5.5 scores eighteen points higher on the same token rates, and at high effort it still scores 47 for $0.88 a task against this row's $5.09.

Inputs

TextImages

API alias

sonnet-5 Names exactly this model version and never moves.

Reasoning effort

lowmediumhighmax We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Twice as cheap as GPT 5.6 Sol per output token and about two and a half times the cost per finished task — a plain case of a rate card not predicting the bill. Streaming speed varies with reasoning effort: this row reports the top measured setting, while MindsHub defaults to high.

Coding specialist for repo-scale changes

Pick it when

The job is software, across many files, and you want a model tuned for that rather than for general reasoning.

Skip it when

The work is not code.

Inputs

TextImages

API alias

gpt-codex-5-3 gpt-codex The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhighxhigh We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to xhigh. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Artificial Analysis scores this version only provisionally — on an older component set that omits both of the benchmarks they run on every fully scored row (Terminal-Bench v4.0 and GDPval-AA v2) — and publishes no cost per task for it, as they do for none of their provisional scores. Their figure is not on the same scale as the rest of this column, so both published columns here are ours.

The previous Luna, for high-volume work already validated on it

Pick it when

You run routine, high-volume work already validated on this version, and re-testing it on GPT-6 Luna is not yet worth the interruption.

Skip it when

You are starting fresh — GPT-6 Luna scores a point higher for $0.07 a task against $0.18, at lower token rates.

Inputs

TextImages

API alias

gpt-luna-5-6 Names exactly this model version and never moves.

Reasoning effort

nonelowmediumhighxhighmax We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to max. Our default can differ from the benchmark in quality, cost and wait.

A MindsHub model alias optimized for speed, served by Cerebras

Pick it when

You want fast generation for interactive work in MindsHub Agents or through the API.

Skip it when

You need to pin a model version or require verified first-answer latency for your workload.

Inputs

TextImages

API alias

mindshub_blaze A rotating alias: we move it as better options ship, so the model behind it can change without a code change on your side.

Reasoning effort

nonehigh We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Served by

Cerebras We buy the inference from Cerebras, which is the name the rate card shows.

Speed source

Cerebras estimate. Provider-published approximate throughput, not a measured MindsHub median.

Worth knowing

We keep Blaze pointed at a model optimized for speed, so the model behind the alias can change. Quality, SciCode and task cost are reference benchmarks, not an independent Blaze evaluation or a forecast of your MindsHub bill. Their source settings may differ from this route. Speed is a Cerebras estimate; first-answer latency has not been verified. The context window shown is the serving route's limit (131,072 tokens on the paid Cerebras tier).

Every input type, at an open-model rate

Pick it when

You need mixed-media input in an existing Muse Spark 1.1 workflow.

Skip it when

You can use a newer Muse: 1.2 scores six points higher here at a lower measured task cost.

Inputs

TextImagesAudioVideo

API alias

muse-spark-1-1 Names exactly this model version and never moves.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Fast, high-volume everyday tasks

Pick it when

You already run on Gemini 3.6 Flash and want to keep its alias while you evaluate a newer Flash.

Skip it when

3.7 Flash scores five index points better, is stronger at code, and finishes a benchmark task for $0.93 against $1.60 on the same token rates.

Inputs

TextImagesAudioVideo

API alias

gemini-flash-3-6 Names exactly this model version and never moves.

Reasoning effort

minimallowmediumhigh We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. Our default can differ from the benchmark in quality, cost and wait.

The legacy Google Flash, for workloads already built on it

Pick it when

You already run on Gemini 3.5 Flash and want the same model behind the same alias while you evaluate a newer Flash.

Skip it when

You are starting fresh: Google itself files this as its legacy Flash, and the newer Flash generations here score higher at a lower rate card.

Inputs

TextImagesAudioVideo

API alias

gemini-flash-3-5 Names exactly this model version and never moves.

Reasoning effort

minimallowmediumhigh We send medium unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. Our default can differ from the benchmark in quality, cost and wait.

Worth knowing

Figures are at high, the top of its dial. Artificial Analysis scores its medium and minimal runs only provisionally on the current index and publishes no cost per task for them, so high is the one comparable reading and the one this row quotes.

Huge documents, images, audio and video

Pick it when

A single request has to carry a whole corpus of mixed media, and it has to be reasoned over rather than skimmed.

Skip it when

You are paying by the token — the Flash line reads the same inputs at a third of the rate card, and 3.7 Flash outscores this by nine points.

Inputs

TextImagesAudioVideo

API alias

gemini-3-1-pro gemini The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

lowmediumhigh We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Preview-era Flash, kept so early integrations keep running

Pick it when

A workload already built on the Gemini 3 preview that you are not ready to move, and want to keep calling by its own alias.

Skip it when

Anything new: it is a preview from December 2025, and each Flash release since scores higher here.

Inputs

TextImagesAudioVideo

API alias

gemini-flash-3 Names exactly this model version and never moves.

Reasoning effort

minimallowmediumhigh We send high unless you ask for another. Benchmark figures use the highest published source setting; this route supports up to high. A lower setting may reduce cost and wait; test whether it still clears your tasks.

Worth knowing

Artificial Analysis lists it in reasoning and non-reasoning configurations and publishes no cost per task for either — on their table, the mark of a score not yet on the current component set — so quality remains marked as provisional and cost per task stays blank. Through MindsHub it exposes a minimal-to-high dial.

Small and cheap for routine automation

Pick it when

Classification, extraction, tagging and other high-frequency jobs with a narrow definition of correct.

Skip it when

Luna passes the same tasks: its index is higher here, with a lower measured task cost. Compare both on your workload.

Inputs

TextImages

API alias

gpt-mini-5-4 gpt-mini The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

nonelowmediumhighxhigh We send none unless you ask for another — the lowest setting this route exposes. Benchmark figures use the highest published source setting; this route supports up to xhigh. Our default can differ from the benchmark in quality, cost and wait.

Built for throughput on bulk, simple work

Pick it when

Millions of tiny calls where throughput matters and the task fits in one sentence.

Skip it when

Any judgement is required.

Inputs

TextImages

API alias

gpt-nano-5-4 gpt-nano The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

nonelowmediumhighxhigh We send none unless you ask for another — the lowest setting this route exposes. Benchmark figures use the highest published source setting; this route supports up to xhigh. Our default can differ from the benchmark in quality, cost and wait.

The smallest Claude, for high-volume simple work

Pick it when

You want Claude behaviour and Claude tooling on work that does not need Claude-sized thinking.

Skip it when

The job needs reasoning headroom — this is the previous generation, and it shows on hard tasks.

Inputs

TextImages

API alias

haiku-4-5 haiku The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Lite-tier extraction and classification at scale

Pick it when

High-volume, simple work — extraction, classification, routing, transcription — where per-call cost and latency matter more than depth.

Skip it when

Multi-step reasoning or coding — the Flash and Pro lines exist for that.

Inputs

TextImagesAudioVideo

API alias

gemini-flash-lite-3-1 gemini-flash-lite The first names exactly this model and never moves. The second is the rotating alias for its line: it points here today and follows the line when we move it. Both reach the same model — see Rotating aliases.

Reasoning effort

minimallowmediumhigh We send minimal unless you ask for another — the lowest setting this route exposes. Benchmark figures use the highest published source setting; this route supports up to high. Our default can differ from the benchmark in quality, cost and wait.

Open weights for focused text tasks

Pick it when

You need downloadable weights or a low-cost text model and can validate the result on your own tasks.

Skip it when

You need image input or frontier-level general reasoning.

Inputs

Text

API alias

gpt-oss-fireworks Names exactly this model version and never moves.

Reasoning effort

Single setting — no dial. This route exposes no effort dial. The benchmark uses the source configuration and endpoint; your wait and cost can differ.

Licence

Apache 2.0

Served by

Fireworks AI OpenAI makes it. We buy the inference from Fireworks AI, which is the name the rate card shows.

Worth knowing

Benchmark figures describe Artificial Analysis’s gpt-oss-120b high-effort configuration and source endpoints, not a measurement of MindsHub’s Fireworks route. The catalog exposes no effort dial for this route.