Models workbench

Price, capability, and service constraints in one working surface. Build a two-to-four model comparison from published records.

Benchmark evidence · Stale · checked Sep 14, 2026

Benchmark revision benchmark_2f59d4658a789000c14f0d0c7a177e62 · cache benchmark_2f59d4658a789000c14f0d0c7a177e62+cache-20260914055705000-c21dcf92-e210-43a5-bb34-f9c3e2271701 · pinned catalog catalog_e7f5de78d2108d49f3be10ea_62bbb2e3 · current catalog catalog_e7f5de78d2108d49f3be10ea_62bbb2e3.

Visible models: 100
Frontier set: Loading…
Selection: 0/4

Price–performance frontier

Composite quality against blended cost per one million tokens.

Frontier only Log scale

Loading the complete price–performance publication…

Quick comparison

Select two to four models from the catalog. The selection stays in the ordered models URL parameter across catalog views and filters.

Selected models

0 / 4

Add a model More details

0 of 4 models selected.

Catalog

100 of 5442 matching source-scoped directory records · sorted by rank · profile links use canonical routes. Weekly rank is a benchmark directory order, not usage popularity.

This is a bounded directory page. More published records are available. Use the page controls; search and access filters query the complete published directory. Your selected models remain in the URL.

Anthropic · Proprietary

Claude Mythos 5

Score: 83.40
Runtime: Not reported
Input / 1M: $10
Output / 1M: $50
Context: 1M
Evidence: supported

Anthropic · Proprietary

Claude Fable 5

Score: 83.15
Runtime: Not reported
Input / 1M: $10
Output / 1M: $50
Context: 1M
Evidence: supported

Anthropic · Proprietary

Claude Opus 5

Score: 83.06
Runtime: Not reported
Input / 1M: $5
Output / 1M: $25
Context: 1M
Evidence: supported

OpenAI · Proprietary

GPT-5.6 Sol

Score: 82.20
Runtime: Not reported
Input / 1M: $5
Output / 1M: $30
Context: 1.1M
Evidence: supported

Moonshot AI · Unknown

Kimi K3

Score: 80.61
Runtime: Not reported
Input / 1M: $3
Output / 1M: $15
Context: 1.1M
Evidence: supported

Alibaba · Open weights

Qwen3.8 Max

Score: 79.22
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Tencent · Open weights

Hy4 preview

Score: 79.16
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Meta · Proprietary

Muse Spark 1.1

Score: 77.14
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Anthropic · Proprietary

Claude Opus 4.8

Score: 76.60
Runtime: Not reported
Input / 1M: $5
Output / 1M: $25
Context: 1M
Evidence: supported

Google · Proprietary

Gemini 3.6 Flash

Score: 75.66
Runtime: Not reported
Input / 1M: $1.5
Output / 1M: $7.5
Context: 1M
Evidence: supported

xAI · Proprietary

Grok 4.5

Score: 75.65
Runtime: Not reported
Input / 1M: $2
Output / 1M: $6
Context: 500K
Evidence: supported

OpenAI · Proprietary

GPT-5.4

Score: 73.56
Runtime: Not reported
Input / 1M: $2.5
Output / 1M: $15
Context: 1.1M
Evidence: supported

OpenAI · Proprietary

GPT-5.6 Terra

Score: 72.95
Runtime: Not reported
Input / 1M: $2.5
Output / 1M: $15
Context: 1.1M
Evidence: estimated

OpenAI · Proprietary

GPT-5.5

Score: 72.92
Runtime: Not reported
Input / 1M: $5
Output / 1M: $30
Context: 1M
Evidence: estimated

Anthropic · Proprietary

Claude Opus 4.7 (Adaptive)

Score: 72.58
Runtime: Not reported
Input / 1M: $5
Output / 1M: $25
Context: 1M
Evidence: estimated

Alibaba · Open weights

Qwen3.8-27B

Score: 72.51
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Anthropic · Proprietary

Claude Opus 4.7

Score: 72.33
Runtime: Not reported
Input / 1M: $5
Output / 1M: $25
Context: 1M
Evidence: supported

Alibaba · Proprietary

Qwen3.7 Max

Score: 71.79
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Meta · Proprietary

Muse Spark

Score: 71.02
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Xiaomi · Proprietary

MiMo-V2.5-Pro

Score: 69.41
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

MiniMax · Open weights

MiniMax M3

Score: 68.73
Runtime: Not reported
Input / 1M: $0.3
Output / 1M: $1.2
Context: 1M
Evidence: supported

Dots Studio · Open weights

dots3-note Preview

Score: 68.66
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Ornith AI · Open weights

Ornith-1.5-397B

Score: 68.60
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Anthropic · Proprietary

Claude Opus 4.6

Score: 68.26
Runtime: Not reported
Input / 1M: $5
Output / 1M: $25
Context: 1M
Evidence: supported

Tencent · Open weights

Hy3

Score: 68.16
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Google · Proprietary

Gemini 3 Pro

Score: 67.64
Runtime: Not reported
Input / 1M: $2
Output / 1M: $12
Context: 2M
Evidence: supported

OpenAI · Proprietary

GPT-5.6 Luna

Score: 67.35
Runtime: Not reported
Input / 1M: $1
Output / 1M: $6
Context: 1.1M
Evidence: estimated

OpenAI · Proprietary

GPT-5.2 Pro

Score: 67.30
Runtime: Not reported
Input / 1M: $21
Output / 1M: $168
Context: 400K
Evidence: supported

Xiaomi · Proprietary

MiMo-V2-Pro

Score: 67.15
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Z.AI · Open weights

GLM-5.1

Score: 67.04
Runtime: Not reported
Input / 1M: $1.4
Output / 1M: $4.4
Context: 203K
Evidence: supported

Thinking Machines Lab · Open weights

Inkling

Score: 67.02
Runtime: Not reported
Input / 1M: $1.87
Output / 1M: $4.68
Context: 1M
Evidence: supported

OpenAI · Proprietary

GPT-5.4 nano

Score: 66.78
Runtime: Not reported
Input / 1M: $0.2
Output / 1M: $1.25
Context: 400K
Evidence: supported

Z.AI · Proprietary

GLM-5-Turbo

Score: 66.30
Runtime: Not reported
Input / 1M: $1.2
Output / 1M: $4
Context: 200K
Evidence: supported

OpenAI · Proprietary

GPT-5.3 Codex

Score: 66.16
Runtime: Not reported
Input / 1M: $1.75
Output / 1M: $14
Context: 400K
Evidence: supported

Alibaba · Proprietary

Qwen3.7 Plus

Score: 65.93
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Z.AI · Open weights

GLM-5

Score: 65.89
Runtime: Not reported
Input / 1M: $1
Output / 1M: $3.2
Context: 200K
Evidence: supported

Google · Proprietary

Gemini 3.5 Flash-Lite

Score: 65.50
Runtime: Not reported
Input / 1M: $0.3
Output / 1M: $2.5
Context: 1M
Evidence: supported

Anthropic · Proprietary

Claude Opus 4.6 (Adaptive)

Score: 65.06
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Anthropic · Proprietary

Claude Sonnet 5

Score: 65.02
Runtime: Not reported
Input / 1M: $2
Output / 1M: $10
Context: 1M
Evidence: estimated

Alibaba · Proprietary

Qwen3.6 Plus

Score: 64.99
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Anthropic · Proprietary

Claude Sonnet 4.6

Score: 64.77
Runtime: Not reported
Input / 1M: $3
Output / 1M: $15
Context: 200K
Evidence: supported

Google · Proprietary

Gemini 3.5 Flash

Score: 64.73
Runtime: Not reported
Input / 1M: $1.5
Output / 1M: $9
Context: 1M
Evidence: estimated

OpenAI · Proprietary

GPT-5.5 Pro

Score: 64.43
Runtime: Not reported
Input / 1M: $30
Output / 1M: $180
Context: 1M
Evidence: estimated

xAI · Proprietary

Grok 4.3

Score: 64.25
Runtime: Not reported
Input / 1M: $1.25
Output / 1M: $2.5
Context: 1M
Evidence: supported

Thinking Machines Lab · Open weights

Inkling-Small

Score: 64.02
Runtime: Not reported
Input / 1M: $0.58
Output / 1M: $1.44
Context: 1M
Evidence: supported

Anthropic · Proprietary

Claude Opus 4.5

Score: 63.91
Runtime: Not reported
Input / 1M: $5
Output / 1M: $25
Context: 200K
Evidence: supported

xAI · Proprietary

Grok 4.6

Score: 63.42
Runtime: Not reported
Input / 1M: $2
Output / 1M: $6
Context: 500K
Evidence: estimated

Z.AI · Open weights

GLM-5.2

Score: 63.40
Runtime: Not reported
Input / 1M: $1.4
Output / 1M: $4.4
Context: 1M
Evidence: estimated

MiniMax · Open weights

MiniMax M2.7

Score: 63.29
Runtime: Not reported
Input / 1M: $0.3
Output / 1M: $1.2
Context: 200K
Evidence: supported

Z.AI · Open weights

GLM-5.3

Score: 62.84
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Xiaomi · Proprietary

MiMo-V2-Omni

Score: 62.61
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Z.AI · Proprietary

GLM-5V-Turbo

Score: 62.53
Runtime: Not reported
Input / 1M: $1.2
Output / 1M: $4
Context: 200K
Evidence: supported

Google · Proprietary

Gemini 3 Pro Deep Think

Score: 62.13
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Meta · Proprietary

Muse Spark 1.2

Score: 61.71
Runtime: Not reported
Input / 1M: $1.25
Output / 1M: $4.25
Context: 1M
Evidence: estimated

OpenAI · Proprietary

GPT-5.4 Pro

Score: 61.53
Runtime: Not reported
Input / 1M: $30
Output / 1M: $180
Context: 1.1M
Evidence: estimated

Google · Unknown

Gemini 3.7 Flash

Score: 61.40
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Alibaba · Open weights

Qwen3.8-Flash-Next

Score: 61.32
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Z.AI · Open weights

GLM-5.3-Flash

Score: 61.27
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

DeepSeek · Proprietary

DeepSeek V4 Pro 0813

Score: 61.02
Runtime: Not reported
Input / 1M: $0.435
Output / 1M: $0.87
Context: 1M
Evidence: estimated

InternScience · Open weights

Agents-A1

Score: 61.02
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

InternScience · Open weights

Agents-A1-F16-GGUF

Score: 61.02
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

InternScience · Open weights

Agents-A1-FP8

Score: 61.02
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

InternScience · Open weights

Agents-A1-Q4_K_M-GGUF

Score: 61.02
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

InternScience · Open weights

Agents-A1-Q8_0-GGUF

Score: 61.02
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Z.AI · Open weights

GLM-4.7

Score: 60.91
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

xAI · Proprietary

Grok 4.1

Score: 60.75
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Z.AI · Open weights

GLM-5 (Reasoning)

Score: 60.58
Runtime: Not reported
Input / 1M: $1
Output / 1M: $3.2
Context: 200K
Evidence: estimated

Alibaba · Proprietary

Qwen 3.6 Max (preview)

Score: 60.41
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Google · Open weights

Gemma 4 31B

Score: 60.39
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Alibaba · Open weights

Qwen3.5 397B (Reasoning)

Score: 60.30
Runtime: Not reported
Input / 1M: $0.6
Output / 1M: $3.6
Context: 128K
Evidence: estimated

Moonshot AI · Proprietary

Kimi K2.5 (Reasoning)

Score: 60.14
Runtime: Not reported
Input / 1M: $0.6
Output / 1M: $3
Context: 256K
Evidence: estimated

Moonshot AI · Open weights

Kimi K2.6

Score: 60.11
Runtime: Not reported
Input / 1M: $0.95
Output / 1M: $4
Context: 256K
Evidence: estimated

Google · Proprietary

Gemini 3 Flash

Score: 60.09
Runtime: Not reported
Input / 1M: $0.5
Output / 1M: $3
Context: 1M
Evidence: supported

Alibaba · Open weights

Qwen3.5-27B

Score: 60.04
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

xAI · Proprietary

Grok 4.1 Fast (Reasoning)

Score: 59.94
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Alibaba · Open weights

Qwen3.5-122B-A10B

Score: 59.89
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

xAI · Proprietary

Grok 4

Score: 59.84
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

OpenAI · Proprietary

GPT-5.2 Instant

Score: 59.80
Runtime: Not reported
Input / 1M: $1.5
Output / 1M: $6
Context: 128K
Evidence: estimated

OpenAI · Proprietary

GPT-5.3 Instant

Score: 59.69
Runtime: Not reported
Input / 1M: $1.75
Output / 1M: $14
Context: 128K
Evidence: estimated

Xiaomi · Proprietary

MiMo-V2.5

Score: 59.48
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

OpenAI · Proprietary

GPT-5 (high)

Score: 59.42
Runtime: Not reported
Input / 1M: $1.25
Output / 1M: $10
Context: 400K
Evidence: estimated

Moonshot AI · Open weights

Kimi K2.5

Score: 59.14
Runtime: Not reported
Input / 1M: $0.6
Output / 1M: $3
Context: 256K
Evidence: supported

MiniMax · Proprietary

MiniMax M2.5

Score: 58.92
Runtime: Not reported
Input / 1M: $0.3
Output / 1M: $1.2
Context: 128K
Evidence: supported

DeepSeek · Open weights

DeepSeek V3.2 (Thinking)

Score: 58.91
Runtime: Not reported
Input / 1M: $0.55
Output / 1M: $2.19
Context: 128K
Evidence: estimated

Alibaba · Open weights

Qwen3 235B 2507 (Reasoning)

Score: 58.80
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

OpenAI · Proprietary

GPT-5.2

Score: 58.39
Runtime: Not reported
Input / 1M: $1.75
Output / 1M: $14
Context: 400K
Evidence: estimated

OpenAI · Proprietary

GPT-5.2-Codex

Score: 58.36
Runtime: Not reported
Input / 1M: $1.75
Output / 1M: $14
Context: 400K
Evidence: supported

Z.AI · Proprietary

GLM-4.5

Score: 58.35
Runtime: Not reported
Input / 1M: $0.6
Output / 1M: $2.2
Context: 128K
Evidence: estimated

Alibaba · Open weights

Qwen3.5 397B

Score: 57.78
Runtime: Not reported
Input / 1M: $0.6
Output / 1M: $3.6
Context: 128K
Evidence: estimated

OpenAI · Proprietary

GPT-5.3-Codex-Spark

Score: 57.68
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Anthropic · Proprietary

Claude Opus 4.5 Thinking

Score: 57.56
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Google · Open weights

Gemma 4 26B A4B

Score: 57.45
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Anthropic · Proprietary

Claude Haiku 4.5

Score: 57.38
Runtime: Not reported
Input / 1M: $1
Output / 1M: $5
Context: 200K
Evidence: estimated

OpenAI · Proprietary

GPT-5.4 mini

Score: 56.96
Runtime: Not reported
Input / 1M: $0.75
Output / 1M: $4.5
Context: 400K
Evidence: estimated

Google · Proprietary

Gemini 2.5 Pro

Score: 56.87
Runtime: Not reported
Input / 1M: $1.25
Output / 1M: $10
Context: 1M
Evidence: supported

Alibaba · Open weights

Qwen3 235B 2507

Score: 56.77
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: estimated

Arcee AI · Open weights

Trinity-Large-Preview

Score: 56.69
Runtime: Not reported
Input / 1M: $0.25
Output / 1M: $1
Context: 512K
Evidence: estimated

Alibaba · Open weights

Qwen3.5-35B-A3B

Score: 56.38
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Google · Proprietary

Gemini 3.1 Pro

Score: 56.25
Runtime: Not reported
Input / 1M: $2
Output / 1M: $12
Context: 1M
Evidence: estimated

xAI · Proprietary

Grok 4 Fast (Reasoning)

Score: 56.11
Runtime: Not reported
Input / 1M: Not reported
Output / 1M: Not reported
Context: Not reported
Evidence: supported

Lifecycle & retirement watch

Lifecycle notices and provider retirement dates are not reported by the published directory.

No verified retirement notices are available. Do not infer a sunset date from absence or record age.

Recent directory observations

First-seen timestamps are directory observations, not vendor release announcements.

Aug 30, 2026Hy4 previewFirst observed in the published directory.
Aug 30, 2026Qwen3.8-Flash-NextFirst observed in the published directory.
Aug 30, 2026GLM-5.3-FlashFirst observed in the published directory.
Aug 23, 2026Qwen3.8-27BFirst observed in the published directory.