Compare AI models
| Specifications | ||
Context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider. | 1M | 1.05M |
| API pricingUSD per 1M tokens · lower wins · base tier | ||
Input priceWhat you pay for everything you send the model — your question, plus any documents or earlier conversation you include with it. | $0.75 | $2.00 |
Output priceWhat you pay for the text the model writes back. It is normally the dearer half: producing an answer costs more than reading one. | $3.75 | $12.00 |
Cached input priceA reduced rate for text you send over and over. If every request starts with the same instructions or the same document, the provider keeps a copy ready and charges less to read it again. | $0.075 | $0.20 |
Cheapest inputLowest input rate across every provider on OpenRouter, the lab included. The cheapest endpoint may run a quantised build or a shorter context — see "Available from" on the model page. | $0.375Google AI Studio | $1.00OpenAI |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $1.875Google AI Studio | $6.00OpenAI |
| $338Google list price | $1,000OpenAI list price | |
| Benchmarks | ||
BullshitBench v2Nonsense detection — Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better. | 39% | 54% |
Gray Swan IPI · k = 1Prompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. Gray Swan's indirect prompt injection benchmark measures how often such an attack succeeds when the attacker gets a single try. Lower is better. | 7.3% | 5.4% |
Gray Swan IPI · k = 10Prompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 10 tries and counts an attack as successful if any of them works. Lower is better. | 32.2% | 26% |
Gray Swan IPI · k = 15Prompt injection robustness — Attackers hide malicious instructions inside content the AI reads — a web page, an email, a document — and try to hijack what it does. This variant gives the attacker 15 tries and counts an attack as successful if any of them works. Lower is better. | 37.3% | 30.4% |
ARC-AGI-2Abstract reasoning — Puzzle-style tests of abstract reasoning and pattern-finding — the kind of thing people find easy but AIs often struggle with. Higher is better. | 60.4% | 83.9% |
threejsevalCommunity preference (Three.js) — Every model gets the same prompt — "the Eiffel Tower", "a glass fishbowl", "a robot arm picking toys into a box" — and builds a 3D scene in Three.js. Real people then see two scenes side by side, names hidden, and vote for the one they prefer. The votes become a chess-style Elo rating on threejseval.com, averaged across all the prompts. It measures whether the scene looks and moves right to a human eye, not whether the code passes a test. Higher is better. | 1349 | 1441 |
ProgramBenchProgram reconstruction — The AI receives a working program and its documentation, then builds a replacement from scratch without the original source code, internet access or decompilation. The score is the percentage of 200 programs that pass every behavioral test. We record each model's best published mini-SWE-agent result, including higher reasoning efforts where available. Partial test-pass rates and almost-solved programs do not count toward this score. Equal scores share a rank here; the official board also uses partial progress to break ties. Higher is better. | 0.5% | — |
CursorBench 4.0Agentic coding — Cursor's own test of coding agents on ambiguous, multi-file tasks taken from real Cursor sessions — editing, refactoring, investigating a codebase, understanding what the user meant, managing jobs and following a design. Cursor runs each model at several reasoning efforts; each release here carries the score of its best listed effort. Scores aren't comparable with earlier CursorBench versions. Higher is better. | — | 41.3% |
DeepSWE 1.1Agentic coding — Artificial Analysis' independent test of deep, agentic software-engineering work — the AI has to plan and carry out substantial coding tasks end to end. (Version 1.1 of the test.) Higher is better. | 49% | — |
MLE-BenchML engineering — Can the AI do the work of a machine-learning engineer? It competes in real Kaggle competitions — building, training, and tuning models end to end — and the score reflects how well it places. Higher is better. | 63.9% | — |
Frontier-Bench v0.1Agentic computer work — A hard, ever-evolving set of real computer tasks — coding, system administration, data work, and more — that an AI agent has to complete on its own. Run by the Harbor / Laude Institute team as the successor to Terminal-Bench (v0.1 is the first release of the task set). The score is the share of tasks solved. Higher is better. | — | 20.8% |
Terminal-Bench 4.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Version 4.0 recalibrated how much time, CPU and memory each task gets, removed eight tasks and fixed nineteen, so fewer runs fail for reasons that have nothing to do with the model. Scores are not comparable with earlier versions. Higher is better. | — | 21.52% |
Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. | — | 84.3% |
BU BenchBrowser agent — Can the AI drive a real web browser to finish tasks — clicking, filling forms, and navigating sites the way a person would? Run by Browser Use on their BU Bench task set. Higher is better. | 68% | — |
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better. | — | 87.5% |
OSWorld-VerifiedAgentic computer use — Can the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better. | 83% | — |
GDPval-AA v2Knowledge work — economically valuable knowledge work (v2, re-based Elo) | 1421 | — |
| Overview | ||
| Company | OpenAI | |
| Release date | Jul 21 2026 | Jun 26 2026 |
| Access | Closed | Closed |
| Model details | View model | View model |
Other comparisons
Gemini 3.6 FlashvsClaude Haiku 5.5GPT-5.6 TerravsClaude Haiku 5.5Gemini 3.6 FlashvsMuse Spark OpenGPT-5.6 TerravsMuse Spark OpenGemini 3.6 FlashvsGrok 4.7GPT-5.6 TerravsGrok 4.7Gemini 3.6 FlashvsDeepSeek-V4.1-Flash-1017GPT-5.6 TerravsDeepSeek-V4.1-Flash-1017Gemini 3.6 FlashvsMistral Large 4GPT-5.6 TerravsMistral Large 4Gemini 3.6 FlashvsKimi K3GPT-5.6 TerravsKimi K3Frequently asked questions
GPT-5.6 Terra leads Gemini 3.6 Flash on 6 of the 6 benchmarks they both report (BullshitBench v2, Gray Swan IPI, ARC-AGI-2, threejseval). Gemini 3.6 Flash is cheaper on both input and output: $0.75 vs $2.00 per million input tokens, and $3.75 vs $12.00 per million output tokens. Figures are base-tier rates. GPT-5.6 Terra shipped 25 days before Gemini 3.6 Flash, so benchmark comparisons should account for the intervening progress.
Context windows are 1M (Gemini 3.6 Flash) vs 1.05M (GPT-5.6 Terra).
On BullshitBench v2, GPT-5.6 Terra leads at 54% vs Gemini 3.6 Flash at 39%. On Gray Swan IPI · k = 1, GPT-5.6 Terra leads at 5.4% vs Gemini 3.6 Flash at 7.3%. On Gray Swan IPI · k = 10, GPT-5.6 Terra leads at 26% vs Gemini 3.6 Flash at 32.2%. On Gray Swan IPI · k = 15, GPT-5.6 Terra leads at 30.4% vs Gemini 3.6 Flash at 37.3%. On ARC-AGI-2, GPT-5.6 Terra leads at 83.9% vs Gemini 3.6 Flash at 60.4%. On threejseval, GPT-5.6 Terra leads at 1441 vs Gemini 3.6 Flash at 1349.