Patchwire

AI lab · 15 models tracked

Anthropic models

Every Anthropic model released in the last 18 months, with release dates, context windows, API prices and benchmark scores. Newest: Claude Haiku 5.5, 13d ago.

Compare Anthropic models → All labs
Models · 30 days
3 of 15
Latest release
13d ago
Cheapest input / 1M
$0.10 Claude Haiku 5.5
Largest context
1M Claude Haiku 5.5

Release history

Timeline

Anthropic ships a new model about every 31 days. Each dot is a release; click one to open it.

Jul 2025Oct 2025Jan 2026Apr 2026Jul 2026 Claude Opus 4.1 — Aug 5, 2025Claude Sonnet 4.5 — Sep 29, 2025Claude Haiku 4.5 — Oct 15, 2025Claude Opus 4.5 — Nov 24, 2025Claude Opus 4.6 — Feb 4, 2026Claude Sonnet 4.6 — Feb 17, 2026Claude Opus 4.7 — Apr 16, 2026Claude Opus 4.8 — May 27, 2026Claude Fable 5 — Jun 9, 2026Claude Sonnet 5 — Jun 30, 2026Claude Opus 5 — Jul 24, 2026Claude Fable 5.1 — Sep 1, 2026Claude Opus 5.5 — Sep 22, 2026Claude Sonnet 5.5 — Sep 28, 2026Claude Haiku 5.5 — Oct 7, 2026

Lineup

All Anthropic models

ModelReleasedContextInput / 1MOutput / 1MGPQA Diamond
Claude Haiku 5.5anthropic/claude-haiku-5.5
ReasoningVision
 Oct 7, 2026 1M128K out $0.10 $0.50 89.6%
Claude Sonnet 5.5anthropic/claude-sonnet-5.5
ReasoningVision
 Sep 28, 2026 1M128K out $2.00 $10.00 95.6%
Claude Opus 5.5anthropic/claude-opus-5.5
ReasoningVision
 Sep 22, 2026 1M128K out $4.00 $20.00 90.6%
Claude Fable 5.1anthropic/claude-fable-5.1
ReasoningVision
 Sep 1, 2026 1M128K out $10.00 $50.00 —
Claude Opus 5anthropic/claude-opus-5
ReasoningVision
 Jul 24, 2026 1M128K out $5.00 $25.00 93.9%
Claude Sonnet 5anthropic/claude-sonnet-5
ReasoningVision
 Jun 30, 2026 1M128K out $2.00 $10.00 90.5%
Claude Fable 5anthropic/claude-fable-5
ReasoningVision
 Jun 9, 2026 1M128K out $10.00 $50.00 85.9%
Claude Opus 4.8anthropic/claude-opus-4.8
ReasoningVision
 May 27, 2026 1M128K out $5.00 $25.00 91.0%
Claude Opus 4.7anthropic/claude-opus-4.7
ReasoningVision
 Apr 16, 2026 1M128K out $5.00 $25.00 90.1%
Claude Sonnet 4.6anthropic/claude-sonnet-4.6
ReasoningVision
 Feb 17, 2026 1M128K out $3.00 $15.00 87.4%
Claude Opus 4.6anthropic/claude-opus-4.6
ReasoningVision
 Feb 4, 2026 1M128K out $5.00 $25.00 90.5%
Claude Opus 4.5anthropic/claude-opus-4.5
ReasoningVision
 Nov 24, 2025 200K64K out $5.00 $25.00 86.1%
Claude Haiku 4.5anthropic/claude-haiku-4.5
ReasoningVision
 Oct 15, 2025 200K64K out $1.00 $5.00 71.2%
Claude Sonnet 4.5anthropic/claude-sonnet-4.5
ReasoningVision
 Sep 29, 2025 1M64K out $3.00 $15.00 82.3%
Claude Opus 4.1anthropic/claude-opus-4.1
ReasoningVision
 Aug 5, 2025 200K32K out $15.00 $75.00 77.3%

Benchmarks

Anthropic's best scores

The top Anthropic model on each benchmark, and where it ranks among all 222 tracked models.

BenchmarkBest Anthropic modelScoreOverall rank
FrontierSWESoftware engineering Claude Opus 5.5 62.3% #2of 19
DeepSWESoftware engineering Claude Opus 5 73.7% #3of 26
CursorBenchSoftware engineering Claude Opus 5.5 57.8% #1of 16
FrontierCodeSoftware engineering Claude Opus 5.5 54.6% #1of 41
SWE-bench VerifiedSoftware engineering Claude Opus 4.7 83.5% #1of 26
Terminal-BenchSoftware engineering Claude Opus 4.7 80.2% #3of 40
MirrorCodeSoftware engineering Claude Opus 5.5 77.4% #1of 9
GBAEvalSoftware engineering Claude Opus 5 79.6% #1of 23
GSOSoftware engineering Claude Fable 5.1 88.2% #1of 20
SciCodeSoftware engineering Claude Opus 5.5 66.9% #1of 93
WeirdML v3Software engineering Claude Opus 5.5 31.2% #3of 22
METR time horizonSoftware engineering Claude Opus 4.6 12 h #1of 13
APEX-AgentsAgents & computer use Claude Sonnet 5.5 75.5% #2of 51
Remote Labor IndexAgents & computer use Claude Opus 5.5 21.3% #1of 13
OSWorld 2.0Agents & computer use Claude Opus 5 31.4% #1of 9
OSWorldAgents & computer use Claude Sonnet 4.6 72.1% #1of 4
GDPvalAgents & computer use Claude Opus 4.5 45.5% #2of 7
Vending-Bench 2Agents & computer use Claude Opus 5 $11,182 #4of 61
PostTrainBenchAgents & computer use Claude Fable 5 41.8% #1of 11
Lech Mazur WritingWriting & content Claude Opus 4.1 8.47/10 #2of 13
DeepResearch BenchWriting & content Claude Opus 4.6 55.3% #1of 20
GDP.pdfWriting & content Claude Sonnet 5.5 31.4% #3of 38
SimpleQA VerifiedWriting & content Claude Opus 5.5 72.2% #5of 66
WebDev ArenaDesign & visual Claude Opus 5.5 1820 #1of 88
Blueprint-Bench 2Design & visual Claude Opus 5.5 51.2% #2of 32
Furniture AssemblyDesign & visual Claude Opus 5.5 83.3% #1of 34
VPCTDesign & visual Claude Opus 4.5 40.0% #6of 9
GeoBenchDesign & visual Claude Opus 4.5 75.0% #3of 3
CybenchCybersecurity Claude Opus 4.6 93.0% #1of 4
ExploitBenchCybersecurity Claude Opus 4.7 26.5% #2of 8
GPQA DiamondReasoning & knowledge Claude Sonnet 5.5 95.6% #3of 92
Humanity's Last ExamReasoning & knowledge Claude Fable 5.1 46.5% #2of 23
ARC-AGI-2Reasoning & knowledge Claude Opus 5.5 93.3% #3of 63
ARC-AGI-1Reasoning & knowledge Claude Opus 5.5 98.5% #1of 63
SimpleBenchReasoning & knowledge Claude Fable 5 81.9% #1of 43
CritPtReasoning & knowledge Claude Opus 5.5 31.7% #2of 96
Chess PuzzlesReasoning & knowledge Claude Opus 5.5 67.0% #4of 70
LMCAReasoning & knowledge Claude Sonnet 5.5 68.3% #1of 96
FrontierMathMaths Claude Opus 5.5 91.2% #3of 72
FrontierMath Tier 4Maths Claude Opus 5.5 95.0% #3of 62
ProofBenchMaths Claude Sonnet 5.5 100.0% #1of 70
OTIS Mock AIMEMaths Claude Sonnet 5.5 100.0% #1of 90

Directory

Other labs