Anthropic · proprietary on Arena · first on Arena 3 October 2025

claude-sonnet-4-5-20250929-thinking-32k

4 days in first place

It held first place on Arena in WebDev (4 days).

Embed

Credits LMArena.

Credits LMArena.

Credits Epoch AI.

Each frame links back to its page. Preview it.

Cite
Download
Post on X

Text: best place 4

last on the board 11 August 2026
1st10th
On the board
3 October 2025 to 11 August 2026
Score
1433 (1431–1436) on 11 August 2026 · 82,549 votes

WebDev: 4 days in first place

last on the board 11 August 2026
1st10th
On the board
12 November 2025 to 11 August 2026
Score
1392 (1384–1400) on 11 August 2026 · 12,803 votes
First place
  • 12 November 2025 to 16 November 2025 · 4 days
Catch-up
Open weights matched it on 22 December 2025, 40 days after it first took first place: glm-4.7.

Epoch AI: Claude Sonnet 4.5

released 29 September 2025 · API access
  1. Chess Puzzles 10.0%
  2. EBR-bench 2.4%
  3. FrontierMath-2025-02-28-Private 15.2%
  4. FrontierMath-Tier-4-2025-07-01-Private 4.2%
  5. FrontierMath-Tier-4-v2-Private 2.4%
  6. FrontierMath-Tiers-1-3-v2-Private 23.9%
  7. GPQA diamond 82.3%
  8. MATH level 5 97.7%
  9. Mystery Game Puzzles 17.0%
  10. OTIS Mock AIME 2024-2025 77.8%
  11. SWE-Bench verified 71.3%
  12. SimpleQA Verified 30.7%

When each vendor retires it

every model’s dates
  • Anthropic 30 Nov 2026 claude-sonnet-4-5-20250929 · 61 days’ notice, from the deprecation on 30 Sep 2026
  • Amazon Bedrock 8 Apr 2027 anthropic.claude-sonnet-4-5-20250929-v1:0 · 182 days’ notice, from the deprecation on 8 Oct 2026

Sources

Arena scores come from people voting between two anonymous answers. First place is the highest score; the top two are often within each other’s margin of error. LMArena, Arena leaderboard dataset, revision 4491b06, retrieved 30 October 2026; Epoch AI, Capabilities & benchmarking, retrieved 30 October 2026. Both CC BY 4.0. Retirement dates: each vendor’s own deprecation page, quoted and kept, listed with the record.