{"schema":"postcutoff/model@1","as_of":"2026-10-30T13:42:00+01:00","url":"https://postcutoff.com/m/gemini-4-carbon/","md":"https://postcutoff.com/m/gemini-4-carbon/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"gemini-4-carbon","name":"Gemini 4 Carbon","org":"Google DeepMind","family":"Gemini 4","released":"2026-10-28","status":"preview","type":"reasoning-llm","modality_in":["text","image","video","audio"],"modality_out":["text"],"open_weights":false,"model_license":"proprietary","context_window":1000000,"max_output":null,"knowledge_cutoff":null,"pricing":{"input":2,"output":10,"unit":"per 1M tokens, introductory price announced by Google (cached input 95% off); after the introductory period $4.00 input / $20.00 output. No end date given. Not yet purchasable by the public (Fairwind partners only at launch)","source":"https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/"},"price_line":"$2 in, $10 out per 1M tokens","access":[{"provider":"Google DeepMind Fairwind Program (a subset of cyber defenders from the week of 2026-10-28, then more trusted partners; public model id not published)","url":"https://deepmind.google/fairwind-program/"},{"provider":"Gemini API / AI Studio, Gemini app, Vertex / Gemini Enterprise Agent Platform: not yet available; Google says 'soon' with no date","url":"https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/"}],"capabilities":[{"name":"Top overall independent score at launch","detail":"Artificial Analysis Intelligence Index v5: 59 (high), #1 of 52 on 2026-10-28, ahead of Claude Opus 5.5 at max (58) and GPT-6 Astra at max (57); #1 on HLE-Diamond, Terminal-Bench-Science, AA-Omniscience and AutomationBench-AA. Second on AA's Coding Agent Index (59 vs Opus 5.5 in Claude Code 62).","first":false,"discovered":"launch","source":"https://x.com/ArtificialAnlys/status/2115494553780367384"},{"name":"Cyber defense lifecycle (find, exploit, patch)","detail":"Vendor-reported: SEC-bench Pro 88.2% (finding), ExploitGym 54.8% (exploitation; GPT-6 Astra 42.4%, Opus 5.5 34.5%), CWE-bench 71.0% (patching, Collinear leaderboard); more than 2x uplift in secure-code generation over Argon on internal tests. Self-computed with mitigations off.","first":false,"discovered":"launch","source":"https://storage.googleapis.com/deepmind-media/gemini/gemini_4_carbon_model_evaluation.pdf"},{"name":"Long-context reasoning","detail":"Vendor-reported GraphWalks BFS 256K-1M F1 96.2% (Argon 84.2%, GPT-6 Astra 71.8%, Opus 5.5 66.8%).","first":false,"discovered":"launch","source":"https://x.com/OfficialLoganK/status/2115489284404940898"},{"name":"Low hallucination rate with higher accuracy","detail":"Artificial Analysis AA-Omniscience v1.1: score 54 (Opus 5.5 max 48), accuracy 58%, hallucination rate 11% (Argon 4%; Opus 5.5 max and GPT-6 Astra max 50%).","first":false,"discovered":"launch","source":"https://x.com/ArtificialAnlys/status/2115494553780367384"},{"name":"Top of Arena's crowd-sourced Text Arena at launch","detail":"Arena (ex-LMArena), Oct 28: #1 in Text Arena (1528, Argon 1520), #2 in Agent Arena ($1.80 median cost per task, 72% below #1 Claude Opus 5.5 Max), #4 in Code Arena WebDev (1768).","first":false,"discovered":"launch","source":"https://x.com/arena/status/2115495002998751262"}],"entry":"2026-10-28-gemini-4-carbon","notes":"Successor to Gemini 4 Argon within the Gemini 4 family, announced 2026-10-28 as a gated release (cyber defenders first through the Fairwind Program). No public API id, docs page, model card or knowledge cutoff yet (ai.google.dev/gemini-api/docs/models/gemini-4-carbon and the deepmind.google model-card URL both returned 404 on 2026-10-28 and again on 2026-10-29); the id will probably be gemini-4-carbon, but that is unconfirmed. Context window 1M tokens and inputs text, image, video and speech per Artificial Analysis. AA measures ~100 output tokens/s and $7.76 per Intelligence Index task at the $4/$20 standard price; Vals AI: Vals Index 70.36% (#2, behind Claude Sonnet 5.5), Terminal-Bench 4.0 63.13%. Until it is generally available, agents should use gemini-4-argon.","verified":"2026-10-29","body_md":"Google's second Gemini 4 model. Not callable by the public at launch; this file will get the API id, endpoints, limits and cutoff\nwhen Google publishes them.\n\nBenchmark table from Google (Logan Kilpatrick's launch post; methodology in the\n[evaluation PDF](https://storage.googleapis.com/deepmind-media/gemini/gemini_4_carbon_model_evaluation.pdf)). Rival scores are mostly\nthe labs' self-reported numbers.\n\n| Benchmark | Carbon | Argon | GPT-6 Astra | Opus 5.5 | Fable 5.1 |\n|---|---|---|---|---|---|\n| AA Index v5.0 | 59 | 50 | 57 | 58 | 49 |\n| Vals Index v2.2 | 70.4 | 69.2 | 65.0 | 70.2 | — |\n| SWE-sweep | 9.6 | 4.3 | 9.5 | 3.6 | 2.9 |\n| FrontierSWE 2 | 61.2 | 55.0 | 65.5 | 62.3 | 56.3 |\n| Terminal-Bench 4.0 (AA) | 65.2 | 57.1 | 59.1 | 59.6 | 52.0 |\n| Agents' Last Exam v1.1 | 77.4 | 69.0 | 73.1 | 71.6 | — |\n| OSWorld-2.1 (offline) | 81.5 | 74.7 | 80.9 | 87.2 | 76.8 |\n| AutomationBench | 53.9 | 51.3 | 41.4 | 42.5 | 31.4 |\n| Vending-Bench 2 ($) | 16,484 | 13,718 | 15,515 | 9,235 | 5,422 |\n| Chartography | 72.5 | 71.6 | 71.0 | 66.3 | 46.2 |\n| Blueprint-Bench 2 | 58.6 | 54.4 | 49.7 | 51.2 | 41.9 |\n| GraphWalks 256K-1M | 96.2 | 84.2 | 71.8 | 66.8 | 65.0 |\n| RiemannBench | 80.0 | 76.0 | 72.0 | 69.6 | 65.6 |\n| Terminal-Bench-Science 0.1 (AA) | 64.8 | 52.9 | 63.3 | 59.0 | 43.3 |\n| CWE-bench | 71.0 | 68.0 | 68.0 | 67.0 | 58.0 |\n| ExploitGym | 54.8 | 42.0 | 42.4 | 34.5 | — |\n| SEC-bench Pro | 88.2 | 83.4 | 85.4 | — | — |","page_url":"https://postcutoff.com/m/gemini-4-carbon/","events_after":null,"major_after":null,"historic_after":null,"missing_at_launch":null,"events_since_release":55,"you_url":"https://postcutoff.com/you/gemini-4-carbon/","briefings":null}