Context News

Available in Claude Code at launch

Will Sonnet 5 be available on the free Claude tier at launch?

Outperforms Opus 4.5 in WebDev on LMArena

400k+ token context window

50% if significantly better than Sonnet 4.5, YES if significantly better than Opus 4.5 (President's judgement)

Will Claude Sonnet 5 debut in the LM Arena Top 3 overall within 7 days of release?

1M+ token context window

"agent swarm" or other more sophisticated multi agent scheme released/announced with launch.

Releases Feb 5th

Will Sonnet 5 reach #1 overall on LM Arena at any point within 14 days?

Any image generation tool included with the release (nano banana type thing, not gif's)

Sonnet 4.x will be released, not Sonnet 5

1.5M+ token context window

Will Sonnet 5 be #1 overall-no-style-control on LM Arena when first added to the leaderboard

Releases Feb 4th

releases today before 1pm ET (feb 3rd)

Releases today at all (feb 3)

Releases before 2pm EST

Releases before 3pm est

Releases before 4pm EST

Reseases before 1pm Feb 4th

Reseases before 2pm Feb 4th

YES

No model picker

Personalized to you, the user

Trained on synthetic data

Very large context window, or another way to give the model long-term memory

Faster speed(smaller model, architecture improvements, mixture of experts keeping most parameters unused during inference, etc.)

Designed to inference cheaper than GPT-4(price must be lower)

Exact-query of training data or knowledge databases(similar to Google RETRO)

Online learning of specialized agents(finetuning, PEFTs, etc.)

Operates autonomously with instructions from you

capable of online learning

New Neural Network Architecture

Seed option to create the question

(Near) real-time update of training data (data delay is within 6 months and is kept updated)

Audio Modality

Acts autonomously

Has a catchier name

uses some non-AdamW optimizer

>= 50% on TerminalBench (https://www.tbench.ai/leaderboard)

Some variation of NSA (Native Sparse Attention)

>= 73% on SWE-Bench Verified (according to epoch.ai)

>= 2 shared experts

>=25T pretraining tokens

>=52B active parameters

>=1.5T parameters

>=16 active experts

1M+ Context

Some image input (multimodality)

>=512 experts

intra-expert communication

DS-MoE with adaptative expert count

>= 60% on BrowseComp (https://www.kaggle.com/benchmarks/openai/browsecomp)

Gemini 2.5 Pro tier or higher on FictionBench (90.6%+ at 192k)

>= 44% on Humanity's Last Exam (text only) at scale.com leaderboard

DeepSeek reports some results with a full-blown deep research agent, and emphasizes that this is the intended use-mode

Other

2042

2043

2044

2045

2041

2040

2027

2039

2037

2034

2035

2036

2038

2024

2025

2026

2028

2029

2030

2031

2032

NO

NO

YES

NO

YES