OutYet.ai

Latest releaseOpenAINov 3, 2026LLM

GPT-6.1 Terra

Is GPT-6.1 Terra out yet? Yes.

OpenAI’s October 26 announcement describes GPT-6.1 Terra for visual reasoning and computer use, with API access and availability in ChatGPT Work and Codex.

Model pageCompare

Agentic76.7○ BenchLM · #6
Knowledge77.1○ BenchLM · #11

News

All updates

  • model · video · audio@pika_labs

    Pika brings H3 Flex to its creation platform

    Pika says H3 Flex is now available on its platform, with faster-than-realtime generation, exportable audio stems, and output up to 4K. Those capabilities and its claim that it leads DesignArena's open-weight text-to-video board come from Pika; the ranking hasn't been independently checked here.

  • model · text · image · code@Kimi_Moonshot

    Moonshot posts Kimi K3.1 Pro weights to Hugging Face

    Moonshot posted Kimi K3.1 Pro weights to Hugging Face and says it will keep advancing open intelligence. The post doesn’t give details on hardware requirements or independent evaluations, so builders will need to inspect the model card before deciding whether they can run it.

  • model · video · audio@pika_labs

    Pika H3 Flex gets a two-step Turbo LoRA

    Pika says @_akhaliq built a Turbo LoRA for H3 Flex less than a day after launch, with 4-step fast and 8-step HD variants for ComfyUI and Diffusers. Pika also calls H3 Flex the top open-weight model on DesignArena’s text-to-video board; that ranking is the company’s claim.

  • tool · agents · code@MistralAI

    Mistral Large 4 builds a live tracker for satellites and debris

    Mistral says it asked Mistral Large 4 to build a real-time 3D tracker for roughly 19,000 satellites and debris fragments. The company says the model built the app, tested it, and passed verification; those are Mistral’s claims.

  • tool · agents · code@Teknium

    Hermes Agent subscriptions now include a cloud agent and computer

    Teknium says each Hermes Agent subscription now comes with a cloud instance, its own computer, higher usage limits, and tools such as search, browser use, and image generation. He also says Grok Bot has been taken offline because serving costs outpaced revenue.

Recently released

The board
OpenAILLM

GPT-6.1 Terra

Is GPT-6.1 Terra out yet?

Yes

AvailablePublicly available right now - we confirmed access.

Released Nov 3, 2026

OpenAI’s October 26 announcement describes GPT-6.1 Terra for visual reasoning and computer use, with API access and availability in ChatGPT Work and Codex.

Details
Google DeepMind logoGoogle DeepMindLLM

Gemini 3.8 Flash

Is Gemini 3.8 Flash out yet?

Yes

AvailablePublicly available right now - we confirmed access.

Released Oct 21, 2026

Google DeepMind's model card identifies Gemini 3.8 Flash as a released multimodal model for coding, agents, and knowledge work, distributed through the Gemini app, AI Studio, and Gemini API.

Details
Meta logoMetaMultimodal

Muse Spark Open

Is Muse Spark Open out yet?

Yes

AvailablePublicly available right now - we confirmed access.

Released Oct 15, 2026

Meta Superintelligence Labs' open-weight Muse Spark: an 816B-parameter mixture-of-experts with 42B active parameters, a 1M-token context window and native image understanding, released under the Muse Model License with weights on Hugging Face.

Details
Google DeepMind logoGoogle DeepMindLLM

Gemini 4 Argon

Is Gemini 4 Argon out yet?

Yes

AvailablePublicly available right now - we confirmed access.

Released Oct 14, 2026

Google DeepMind's frontier model, announced September 30, 2026 and rolled out October 14 to paying Gemini and Antigravity users and the Gemini API (free tier excluded). One-million-token context window; API launch pricing of $2 per million input tokens and $10 per million output through November 30, then $4 and $20.

Details
Alibaba QwenImage model

Qwen-Image-2.1-Turbo

Is Qwen-Image-2.1-Turbo out yet?

Yes

AvailablePublicly available right now - we confirmed access.

Released Oct 9, 2026

Accelerated checkpoint of Qwen-Image-2.1: the same 7B visual generator rendering and editing in 8 denoising steps instead of 40, with open weights and a hosted Turbo API.

Details

Frontier benchmarks

Compare models

Frontier board

6 releases. Five ways to compare.

Selected for recent releases and benchmark coverage, not a single overall rank. Each column sorts the same models by its own score. No reading means no bar.

○ Independent evaluation · ◆ Vendor or self-reported, labeled per reading.

Coding

BenchLM · 0 to 100

  1. Claude Opus 5.582.87Independent

    Anthropic

  2. Claude Fable 5.176.4Independent

    Anthropic

  3. GPT-6 Astra74.71Independent

    OpenAI

  4. Gemini 4 Argon72.16Independent

    Google DeepMind

  5. GPT-6.1 Sol70.29Independent

    OpenAI

  6. Gemini 3.8 Flash59.8Independent

    Google DeepMind

Agentic

BenchLM · 0 to 100

  1. Claude Opus 5.590.07Independent

    Anthropic

  2. Claude Fable 5.180.7Independent

    Anthropic

  3. Gemini 4 Argon78.26Independent

    Google DeepMind

  4. GPT-6.1 Sol77.24Independent

    OpenAI

  5. GPT-6 Astra75.97Independent

    OpenAI

  6. Gemini 3.8 Flash65.79Independent

    Google DeepMind

Reasoning

BenchLM · 0 to 100

  1. GPT-6 Astra93.3Independent

    OpenAI

  2. GPT-6.1 Sol90.5Independent

    OpenAI

  3. Gemini 4 Argon86.9Independent

    Google DeepMind

  4. Claude Opus 5.586.1Independent

    Anthropic

  5. Claude Fable 5.185.2Independent

    Anthropic

  6. Gemini 3.8 Flash74.3Independent

    Google DeepMind

Knowledge

BenchLM · 0 to 100

  1. Claude Opus 5.589.53Independent

    Anthropic

  2. GPT-6.1 Sol86.18Independent

    OpenAI

  3. Claude Fable 5.186.04Independent

    Anthropic

  4. GPT-6 Astra85.76Independent

    OpenAI

  5. Gemini 4 Argon82.61Independent

    Google DeepMind

  6. Gemini 3.8 Flash69.19Independent

    Google DeepMind

Arena

LM Arena · Elo

  1. Gemini 4 Argon1,517.17Independent

    Google DeepMind

  2. Claude Opus 5.51,508.73Independent

    Anthropic

  3. Claude Fable 5.11,503.19Independent

    Anthropic

  4. Gemini 3.8 Flash1,495.7Independent

    Google DeepMind

  5. GPT-6.1 Sol1,479.94Independent

    OpenAI

  6. GPT-6 Astra1,476.48Independent

    OpenAI

Score links open the source; hover for capture details. Bar lengths start at zero. Arena lengths use the full public board maximum.

The shape of the scores

First three featured models. Toggle a name to isolate a line. Missing readings leave gaps, not zeroes.

CodingAgenticReasoningKnowledgeArena050100Claude Opus 5.5, Coding: 82.87. Independent. Captured 2026-11-04.Claude Opus 5.5, Agentic: 90.07. Independent. Captured 2026-11-04.Claude Opus 5.5, Reasoning: 86.1. Independent. Captured 2026-11-04.Claude Opus 5.5, Knowledge: 89.53. Independent. Captured 2026-11-04.Claude Opus 5.5, Arena: 1,508.73. Independent. Captured 2026-10-26.Claude Fable 5.1, Coding: 76.4. Independent. Captured 2026-11-04.Claude Fable 5.1, Agentic: 80.7. Independent. Captured 2026-11-04.Claude Fable 5.1, Reasoning: 85.2. Independent. Captured 2026-11-04.Claude Fable 5.1, Knowledge: 86.04. Independent. Captured 2026-11-04.Claude Fable 5.1, Arena: 1,503.19. Independent. Captured 2026-10-26.GPT-6 Astra, Coding: 74.71. Independent. Captured 2026-11-04.GPT-6 Astra, Agentic: 75.97. Independent. Captured 2026-11-04.GPT-6 Astra, Reasoning: 93.3. Independent. Captured 2026-11-04.GPT-6 Astra, Knowledge: 85.76. Independent. Captured 2026-11-04.GPT-6 Astra, Arena: 1,476.48. Independent. Captured 2026-10-26.

3 of 3 series shown.

Axes share drawing space, not meaning. Arena is Elo divided by the current board maximum; it is not a percentage accuracy.

The newest arrival

Gemini 3.8 Flash

Released 2026-10-21 · At least two readings

No predecessor is linked in the catalog yet. We do not infer a generation jump from a name.

Latest release with linked lineage

Claude Fable 5.1 vs Claude Fable 5

Coding
+5.03pts
71.37 to 76.4 Independent2026-11-04 / 2026-11-04
Agentic
+5.43pts
75.27 to 80.7 Independent2026-11-04 / 2026-11-04
Knowledge
+4.58pts
81.46 to 86.04 Independent2026-11-04 / 2026-11-04
Arena
-0.09Elo
1,503.28 to 1,503.19 Independent2026-10-26 / 2026-10-26

Latest snapshots, not a controlled before-and-after trial. Capture dates and evaluation setups may differ.

Scores and sources

Frontier board: original scores. Not reported is missing, not zero.
ModelMetricScore / provenanceCapturedSource label / votes
Claude Opus 5.5CodingBenchLM.ai82.87Independent2026-11-04Claude Opus 5.5
Claude Fable 5.1CodingBenchLM.ai76.4Independent2026-11-04Claude Fable 5.1
GPT-6 AstraCodingBenchLM.ai74.71Independent2026-11-04GPT-6 Astra
Gemini 4 ArgonCodingBenchLM.ai72.16Independent2026-11-04Gemini 4 Argon
GPT-6.1 SolCodingBenchLM.ai70.29Independent2026-11-04GPT-6.1 Sol
Gemini 3.8 FlashCodingBenchLM.ai59.8Independent2026-11-04Gemini 3.8 Flash
Claude Opus 5.5AgenticBenchLM.ai90.07Independent2026-11-04Claude Opus 5.5
Claude Fable 5.1AgenticBenchLM.ai80.7Independent2026-11-04Claude Fable 5.1
Gemini 4 ArgonAgenticBenchLM.ai78.26Independent2026-11-04Gemini 4 Argon
GPT-6.1 SolAgenticBenchLM.ai77.24Independent2026-11-04GPT-6.1 Sol
GPT-6 AstraAgenticBenchLM.ai75.97Independent2026-11-04GPT-6 Astra
Gemini 3.8 FlashAgenticBenchLM.ai65.79Independent2026-11-04Gemini 3.8 Flash
GPT-6 AstraReasoningBenchLM.ai93.3Independent2026-11-04GPT-6 Astra
GPT-6.1 SolReasoningBenchLM.ai90.5Independent2026-11-04GPT-6.1 Sol
Gemini 4 ArgonReasoningBenchLM.ai86.9Independent2026-11-04Gemini 4 Argon
Claude Opus 5.5ReasoningBenchLM.ai86.1Independent2026-11-04Claude Opus 5.5
Claude Fable 5.1ReasoningBenchLM.ai85.2Independent2026-11-04Claude Fable 5.1
Gemini 3.8 FlashReasoningBenchLM.ai74.3Independent2026-11-04Gemini 3.8 Flash
Claude Opus 5.5KnowledgeBenchLM.ai89.53Independent2026-11-04Claude Opus 5.5
GPT-6.1 SolKnowledgeBenchLM.ai86.18Independent2026-11-04GPT-6.1 Sol
Claude Fable 5.1KnowledgeBenchLM.ai86.04Independent2026-11-04Claude Fable 5.1
GPT-6 AstraKnowledgeBenchLM.ai85.76Independent2026-11-04GPT-6 Astra
Gemini 4 ArgonKnowledgeBenchLM.ai82.61Independent2026-11-04Gemini 4 Argon
Gemini 3.8 FlashKnowledgeBenchLM.ai69.19Independent2026-11-04Gemini 3.8 Flash
Gemini 4 ArgonArenaLM Arena1,517.17Independent2026-10-26Gemini 4 Argon (High)10,083 votes
Claude Opus 5.5ArenaLM Arena1,508.73Independent2026-10-26Claude Opus 5.5 (High)10,322 votes
Claude Fable 5.1ArenaLM Arena1,503.19Independent2026-10-26Claude Fable 5.1 (Max)16,343 votes
Gemini 3.8 FlashArenaLM Arena1,495.7Independent2026-10-26Gemini 3.8 Flash (High)43,997 votes
GPT-6.1 SolArenaLM Arena1,479.94Independent2026-10-26GPT-6.1 Sol (Max)8,305 votes
GPT-6 AstraArenaLM Arena1,476.48Independent2026-10-26GPT-6 Astra (Max)14,076 votes
Additional readings for these models. Separate benchmarks, not substitutes for the five chart axes.
ModelMetric / sourceScore / provenanceCaptured
Claude Opus 5.5MultimodalGroundedBenchLM.ai91.8Independent2026-11-04
Claude Opus 5.5GDP.pdfEpoch AI30.6Independent2026-11-04
Claude Fable 5.1Humanity's Last ExamEpoch AI46.5Independent2026-11-04
Claude Fable 5.1GDP.pdfEpoch AI29.6Independent2026-11-04
GPT-6 AstraHumanity's Last ExamEpoch AI54.8Independent2026-11-04
Gemini 3.8 FlashMultimodalGroundedBenchLM.ai88.2Independent2026-11-04
Gemini 3.8 FlashHumanity's Last ExamEpoch AI44.5Independent2026-11-04
Gemini 3.8 FlashGDP.pdfEpoch AI23.4Independent2026-11-04