The daily briefing

Gemini 4 Argon: Google expands Gemini 4 Argon access and publishes API pricing

Wednesday, October 14, 2026 · what moved in the AI release cycle in those 24 hours.

One click creates your free account (or signs you into an existing one) and turns on alerts for the Daily Briefing. There's no password to set, and you can unsubscribe any time.

New benchmark readings

Amazon Nova 2 Pro GDP.pdf: 2 · Epoch AI
Amazon Nova Humanity's Last Exam: 3.6 · Epoch AI
Claude 3.5 Sonnet Humanity's Last Exam: 4.1 · Epoch AI
Claude 3.7 Sonnet Humanity's Last Exam: 8 · Epoch AI
Claude Fable 5 Coding: 71.36 · BenchLM.ai
Claude Fable 5 Agentic: 74.82 · BenchLM.ai
Claude Fable 5 GDP.pdf: 30 · Epoch AI
Claude Fable 5 Knowledge: 82 · BenchLM.ai
Claude Fable 5.1 Coding: 77.02 · BenchLM.ai
Claude Fable 5.1 Agentic: 79.24 · BenchLM.ai
Claude Fable 5.1 Reasoning: 85.2 · BenchLM.ai
Claude Fable 5.1 Knowledge: 86.3 · BenchLM.ai
Claude Haiku 4.5 ExploitBench: 13.4 · Epoch AI
Claude Haiku 5.5 Humanity's Last Exam: 30.1 · Epoch AI
Claude Mythos 5 ExploitBench: 70.6 · Epoch AI
Claude Opus 4 Humanity's Last Exam: 6.7 · Epoch AI
Claude Opus 4.1 Humanity's Last Exam: 7.9 · Epoch AI
Claude Opus 4.5 Math: 61.3 · BenchLM.ai
Claude Opus 4.5 MultimodalGrounded: 48.6 · BenchLM.ai
Claude Opus 4.5 Multilingual: 96.5 · BenchLM.ai
Claude Opus 4.5 Humanity's Last Exam: 14.2 · Epoch AI
Claude Opus 4.6 LM Arena: 1497 · LM Arena
Claude Opus 4.6 Knowledge: 59.3 · BenchLM.ai
Claude Opus 4.6 MultimodalGrounded: 74.5 · BenchLM.ai
Claude Opus 4.6 Humanity's Last Exam: 19 · Epoch AI
Claude Opus 4.7 LM Arena: 1494 · LM Arena
Claude Opus 4.7 Coding: 55.57 · BenchLM.ai
Claude Opus 4.7 Agentic: 54 · BenchLM.ai
Claude Opus 4.7 Knowledge: 65.33 · BenchLM.ai
Claude Opus 4.7 Humanity's Last Exam: 36.2 · Epoch AI
Claude Opus 4.7 ExploitBench: 23.9 · Epoch AI
Claude Opus 4.7 GDP.pdf: 21 · Epoch AI
Claude Opus 4.7 MultimodalGrounded: 53.7 · BenchLM.ai
Claude Opus 4.8 Coding: 60.79 · BenchLM.ai
Claude Opus 4.8 Agentic: 62.22 · BenchLM.ai
Claude Opus 4.8 Knowledge: 71.37 · BenchLM.ai
Claude Opus 4.8 MultimodalGrounded: 91.2 · BenchLM.ai
Claude Opus 4.8 GDP.pdf: 24 · Epoch AI
Claude Opus 4.8 Math: 84.4 · BenchLM.ai
Claude Opus 4.8 Reasoning: 62.8 · BenchLM.ai
Claude Opus 5 Coding: 70.86 · BenchLM.ai
Claude Opus 5 Agentic: 77.58 · BenchLM.ai
Claude Opus 5 Knowledge: 81.63 · BenchLM.ai
Claude Opus 5 MultimodalGrounded: 92.2 · BenchLM.ai
Claude Opus 5 Reasoning: 81 · BenchLM.ai
Claude Sonnet 4 Humanity's Last Exam: 5.5 · Epoch AI
Claude Sonnet 4.5 Humanity's Last Exam: 7.5 · Epoch AI
Claude Sonnet 4.6 MultimodalGrounded: 58.7 · BenchLM.ai
Claude Sonnet 4.6 ExploitBench: 23.6 · Epoch AI
Claude Sonnet 5 Coding: 57.75 · BenchLM.ai
Claude Sonnet 5 Agentic: 64.86 · BenchLM.ai
Claude Sonnet 5 Knowledge: 65.11 · BenchLM.ai
Claude Sonnet 5 MultimodalGrounded: 82 · BenchLM.ai
Command A InstructionFollowing: 90.7 · BenchLM.ai
DeepSeek V4 Pro SWE-Bench Pro: 55.4 · Hugging Faceself reported
DeepSeek-V3 Multilingual: 55.5 · BenchLM.ai
Gemini 1.5 Pro Humanity's Last Exam: 4.6 · Epoch AI
Gemini 2.0 Flash Humanity's Last Exam: 6.6 · Epoch AI
Gemini 2.5 Flash Humanity's Last Exam: 11 · Epoch AI
Gemini 2.5 Pro Humanity's Last Exam: 17.8 · Epoch AI
Gemini 3 Pro Knowledge: 60.49 · BenchLM.ai
Gemini 3 Pro MultimodalGrounded: 80 · BenchLM.ai
Gemini 3 Pro InstructionFollowing: 86.1 · BenchLM.ai
Gemini 3 Pro Humanity's Last Exam: 37.5 · Epoch AI
Gemini 3 Pro LM Arena: 1485 · LM Arena
Gemini 3.1 Pro Knowledge: 67.35 · BenchLM.ai
Gemini 3.1 Pro MultimodalGrounded: 80.1 · BenchLM.ai
Gemini 3.1 Pro GDP.pdf: 17 · Epoch AI
Gemini 3.1 Pro Humanity's Last Exam: 46.4 · Epoch AI
Gemini 3.1 Pro ExploitBench: 26.1 · Epoch AI
Gemini 3.1 Pro LM Arena: 1487 · LM Arena
Gemini 3.5 Flash Coding: 50.38 · BenchLM.ai
Gemini 3.5 Flash Agentic: 50.72 · BenchLM.ai
Gemini 3.5 Flash Reasoning: 70.9 · BenchLM.ai
Gemini 3.5 Flash MultimodalGrounded: 87.8 · BenchLM.ai
Gemini 3.5 Flash InstructionFollowing: 87.4 · BenchLM.ai
Gemini 3.5 Flash Knowledge: 65.3 · BenchLM.ai
Gemini 3.6 Flash Coding: 51.12 · BenchLM.ai
Gemini 3.6 Flash Knowledge: 64.06 · BenchLM.ai
Gemini 3.6 Flash GDP.pdf: 14 · Epoch AI
Gemini 3.7 Flash Coding: 57.83 · BenchLM.ai
Gemini 3.7 Flash Agentic: 57.88 · BenchLM.ai
Gemini 3.7 Flash Knowledge: 71.04 · BenchLM.ai
Gemini 3.7 Flash MultimodalGrounded: 87.2 · BenchLM.ai
Gemini 4 Argon Coding: 71.5 · BenchLM.ai
Gemini 4 Argon Agentic: 74.1 · BenchLM.ai
Gemini 4 Argon Knowledge: 78.93 · BenchLM.ai
Glm 5 2 Coding: 53.9 · BenchLM.ai
Glm 5 2 Agentic: 56.5 · BenchLM.ai
Glm 5 2 InstructionFollowing: 89.9 · BenchLM.ai
Glm 5 2 Math: 78.4 · BenchLM.ai
GLM-4.5 Humanity's Last Exam: 8.1 · Epoch AI
GLM-4.6 SWE-Bench Pro: 9.7 · Hugging Faceself reported
GLM-5 Multilingual: 89.5 · BenchLM.ai
GLM-5 InstructionFollowing: 88.6 · BenchLM.ai
GLM-5.1 Coding: 47.51 · BenchLM.ai
GLM-5.1 Math: 66.7 · BenchLM.ai
GLM-5.1 InstructionFollowing: 93.8 · BenchLM.ai
GLM-5.1 ExploitBench: 16 · Epoch AI
GLM-5.1 SWE-Bench Pro: 58.4 · Hugging Faceself reported
GLM-5.2 SWE-Bench Pro: 62.1 · Hugging Faceself reported
GPT-4.1 Humanity's Last Exam: 5.4 · Epoch AI
GPT-4.1 Multilingual: 61.5 · BenchLM.ai
GPT-4.5 Humanity's Last Exam: 5.4 · Epoch AI
GPT-4o Humanity's Last Exam: 2.7 · Epoch AI
GPT-4o Multilingual: 30.2 · BenchLM.ai
GPT-5 InstructionFollowing: 86.4 · BenchLM.ai
GPT-5 Humanity's Last Exam: 19.4 · Epoch AI
GPT-5.1 InstructionFollowing: 89.4 · BenchLM.ai
GPT-5.1 Humanity's Last Exam: 6.8 · Epoch AI
GPT-5.2 Knowledge: 60.16 · BenchLM.ai
GPT-5.2 MultimodalGrounded: 75.9 · BenchLM.ai
GPT-5.2 InstructionFollowing: 92.6 · BenchLM.ai
GPT-5.2 Humanity's Last Exam: 27.8 · Epoch AI
GPT-5.4 Coding: 48.04 · BenchLM.ai
GPT-5.4 Agentic: 51.28 · BenchLM.ai
GPT-5.4 Reasoning: 64.2 · BenchLM.ai
GPT-5.4 Knowledge: 68.84 · BenchLM.ai
GPT-5.4 MultimodalGrounded: 73.1 · BenchLM.ai
GPT-5.4 InstructionFollowing: 90.7 · BenchLM.ai
GPT-5.4 Humanity's Last Exam: 36.2 · Epoch AI
GPT-5.5 Agentic: 61.46 · BenchLM.ai
GPT-5.5 GDP.pdf: 26 · Epoch AI
GPT-5.5 MultimodalGrounded: 75.5 · BenchLM.ai
GPT-5.5 Coding: 62.13 · BenchLM.ai
GPT-5.5 Knowledge: 71.38 · BenchLM.ai
GPT-5.5 InstructionFollowing: 93.3 · BenchLM.ai
GPT-5.5 ExploitBench: 41.8 · Epoch AI
GPT-5.5 Reasoning: 70.4 · BenchLM.ai
GPT-5.6 Luna Coding: 61.81 · BenchLM.ai
GPT-5.6 Luna Agentic: 54.48 · BenchLM.ai
GPT-5.6 Luna Knowledge: 64.48 · BenchLM.ai
GPT-5.6 Luna MultimodalGrounded: 79.6 · BenchLM.ai
GPT-5.6 Luna Reasoning: 58.2 · BenchLM.ai
GPT-5.6 Sol Coding: 67.64 · BenchLM.ai
GPT-5.6 Sol Agentic: 67.38 · BenchLM.ai
GPT-5.6 Sol Knowledge: 76.93 · BenchLM.ai
GPT-5.6 Sol MultimodalGrounded: 89.8 · BenchLM.ai
GPT-5.6 Sol InstructionFollowing: 89.1 · BenchLM.ai
GPT-5.6 Sol GDP.pdf: 30.7 · Epoch AI
GPT-5.6 Sol Reasoning: 76.4 · BenchLM.ai
GPT-5.6 Terra Coding: 62.18 · BenchLM.ai
GPT-5.6 Terra Agentic: 58.87 · BenchLM.ai
GPT-5.6 Terra Knowledge: 69.91 · BenchLM.ai
GPT-5.6 Terra MultimodalGrounded: 84.5 · BenchLM.ai
GPT-5.6 Terra InstructionFollowing: 87.1 · BenchLM.ai
GPT-5.6 Terra Reasoning: 69.1 · BenchLM.ai
GPT-6 Astra Coding: 75.67 · BenchLM.ai
GPT-6 Astra Agentic: 70.64 · BenchLM.ai
GPT-6 Astra Reasoning: 93.3 · BenchLM.ai
GPT-6 Astra Knowledge: 85.83 · BenchLM.ai
GPT-6 Astra Humanity's Last Exam: 54.8 · Epoch AI
GPT-6 Sol Coding: 66.44 · BenchLM.ai
GPT-6 Sol Agentic: 60.13 · BenchLM.ai
GPT-6 Sol Reasoning: 73.2 · BenchLM.ai
GPT-6 Sol Knowledge: 79.44 · BenchLM.ai
GPT-6.1 Sol Coding: 71.06 · BenchLM.ai
GPT-6.1 Sol Agentic: 74.32 · BenchLM.ai
GPT-6.1 Sol Reasoning: 90.5 · BenchLM.ai
GPT-6.1 Sol Knowledge: 85.51 · BenchLM.ai
gpt-oss SWE-Bench Pro: 16.2 · Hugging Faceself reported
gpt-oss InstructionFollowing: 84.3 · BenchLM.ai
Grok 4.20 MultimodalGrounded: 53.4 · BenchLM.ai
Grok 4.3 GDP.pdf: 8 · Epoch AI
Grok 4.3 InstructionFollowing: 97.1 · BenchLM.ai
Grok 4.3 MultimodalGrounded: 78.8 · BenchLM.ai
Grok 4.5 Coding: 56.07 · BenchLM.ai
Grok 4.5 Agentic: 57.51 · BenchLM.ai
Grok 4.5 Knowledge: 67.18 · BenchLM.ai
Grok 4.5 Reasoning: 54.1 · BenchLM.ai
Grok 4.6 Coding: 58.71 · BenchLM.ai
Grok 4.6 Agentic: 68.29 · BenchLM.ai
Grok 4.6 Knowledge: 69.02 · BenchLM.ai
Grok 4.6 Reasoning: 61.4 · BenchLM.ai
Grok 4.7 Reasoning: 78.5 · BenchLM.ai
Kimi K2 SWE-Bench Pro: 27.7 · Hugging Faceself reported
Kimi K2.5 MultimodalGrounded: 69 · BenchLM.ai
Kimi K2.5 Multilingual: 87.3 · BenchLM.ai
Kimi K2.5 InstructionFollowing: 85.8 · BenchLM.ai
Kimi K2.5 Humanity's Last Exam: 24.4 · Epoch AI
Kimi K2.5 SWE-Bench Pro: 50.7 · Hugging Faceself reported
Kimi K2.6 MultimodalGrounded: 73.8 · BenchLM.ai
Kimi K2.6 GDP.pdf: 12 · Epoch AI
Kimi K2.6 ExploitBench: 16.3 · Epoch AI
Kimi K2.6 SWE-Bench Pro: 58.6 · Hugging Faceself reported
Kimi K2.6 Math: 76.8 · BenchLM.ai
Kimi K2.6 Knowledge: 59.95 · BenchLM.ai
Kimi K3 MultimodalGrounded: 92.6 · BenchLM.ai
Kimi K3 Coding: 59.16 · BenchLM.ai
Kimi K3 Agentic: 68.92 · BenchLM.ai
Kimi K3 Knowledge: 68.46 · BenchLM.ai
Kimi K3 Reasoning: 70 · BenchLM.ai
Llama 4 Humanity's Last Exam: 5.7 · Epoch AI
MiniMax M2.1 SWE-Bench Pro: 36.8 · Hugging Faceself reported
MiniMax M2.5 SWE-Bench Pro: 55.4 · Hugging Faceself reported
MiniMax M2.7 InstructionFollowing: 93 · BenchLM.ai
MiniMax M2.7 ExploitBench: 13.3 · Epoch AI
MiniMax M2.7 SWE-Bench Pro: 56.2 · Hugging Faceself reported
MiniMax M3 SWE-Bench Pro: 59 · Hugging Faceself reported
MiniMax M3 MultimodalGrounded: 61.5 · BenchLM.ai
MiniMax M3 InstructionFollowing: 93.8 · BenchLM.ai
MiniMax M3 Reasoning: 82.9 · BenchLM.ai
Mistral Large 3 GDP.pdf: 2 · Epoch AI
Mistral Large 4 Coding: 46.71 · BenchLM.ai
Mistral Large 4 Agentic: 52.63 · BenchLM.ai
Mistral Large 4 Reasoning: 81.7 · BenchLM.ai
Mistral Medium 3 Humanity's Last Exam: 4.5 · Epoch AI
Muse Spark Humanity's Last Exam: 40.6 · Epoch AI
Muse Spark LM Arena: 1489 · LM Arena
Muse Spark Coding: 50.13 · BenchLM.ai
Muse Spark Agentic: 50.39 · BenchLM.ai
Muse Spark MultimodalGrounded: 84.4 · BenchLM.ai
Muse Spark Knowledge: 62.07 · BenchLM.ai
Muse Spark InstructionFollowing: 93.3 · BenchLM.ai
Nemotron 3 Reasoning: 70.7 · BenchLM.ai
Nemotron 3 Multilingual: 89.2 · BenchLM.ai
Nemotron 3 InstructionFollowing: 92.1 · BenchLM.ai
Nemotron 3 GDP.pdf: 2 · Epoch AI
North Mini Code SWE-Bench Pro: 40.2 · Hugging Faceself reported
o1 InstructionFollowing: 86 · BenchLM.ai
o1 Humanity's Last Exam: 8 · Epoch AI
o3 Humanity's Last Exam: 19.2 · Epoch AI
o3 InstructionFollowing: 87.4 · BenchLM.ai
o4-mini Humanity's Last Exam: 14.3 · Epoch AI
Phi-4 Multilingual: 1 · BenchLM.ai
Qwen 3 SWE-Bench Pro: 21.4 · Hugging Faceself reported
Qwen 3 Multilingual: 79.5 · BenchLM.ai
Qwen 3.5 MultimodalGrounded: 64.6 · BenchLM.ai
Qwen 3.5 Multilingual: 83.8 · BenchLM.ai
Qwen 3.5 InstructionFollowing: 88.8 · BenchLM.ai
Qwen 3.6 MultimodalGrounded: 66.4 · BenchLM.ai
Qwen 3.6 Multilingual: 93.8 · BenchLM.ai
Qwen 3.6 InstructionFollowing: 85.8 · BenchLM.ai
Qwen 3.6 SWE-Bench Pro: 49.5 · Hugging Faceself reported
Qwen 3.6 Math: 58.4 · BenchLM.ai
Qwen 3.7 Max Knowledge: 61.07 · BenchLM.ai
Qwen 3.7 Max Multilingual: 100 · BenchLM.ai
Qwen 3.7 Max InstructionFollowing: 92.8 · BenchLM.ai
Qwen 3.7 Max Math: 79 · BenchLM.ai
Qwen3-Coder SWE-Bench Pro: 38.7 · Hugging Faceself reported
Qwen3-Coder-Next SWE-Bench Pro: 44.3 · Hugging Faceself reported

News & rumors added

NewsGemini 4 Argon

Google expands Gemini 4 Argon access and publishes API pricing

Google says Gemini 4 Argon access is expanding across its products, including the Gemini app and developer APIs. The API launch price is $2 per million input tokens and $10 per…

Google · Oct 13, 2026

Read at blog.google

NewsGemini 4 Argon

Google publishes Gemini 4 Argon safety and security evaluations

Google’s Gemini 4 Argon model card reports that the company could not rule out the model reaching its biological and cybersecurity critical capability thresholds, or a stealth a…

Google DeepMind · Oct 13, 2026

Read at deepmind.google Read OutYet report

Every story we’re tracking →