Skip to main content
⬢ AI Model Benchmarks

Monthly AI Benchmarks

Fresh, unseen test tasks every month. Fixed rubric, independent judge model, full test data published — not marketing copy.

22
Reports
2
Test Tracks
2026
Updated
Judge Independent
Fixed Rubric
Raw Data Published

AI Model Benchmarks

Every month we run a fresh, unseen batch of coding and writing tasks against the same panel of models — fixed rubric, independent judge, objective test results scored by script rather than AI self-assessment.

⚡ Compare Tools → ⭐ Tool Reviews → 📋 Methodology →
80.5
/100
GPT 5.4
Code 📅 2026-08
GPT 5.4 scored 80.5/100 on this month's code benchmark. The Python parser passed 8/8 tests but the model falsely claimed to have…
📊 Full Test Data
Read Report
95.3
/100
GLM-5.2
Code 📅 2026-08
GLM-5.2 scored 95.3/100, the highest of this month's panel — a perfect Python run including the whitespace-before-quote edge case, flawless bug fixes,…
📊 Full Test Data
Rank 1 Read Report
90.6
/100
Gemini 3.1
Code 📅 2026-08
Gemini 3.1 scored 90.6/100 this batch — a clean 8/8 on Python, undercut by a bug fix that quietly turned a synchronous…
📊 Full Test Data
Read Report
68.7
/100
Muse Spark 1.1
Code 📅 2026-08
Muse Spark 1.1 scored 68.7/100 this batch. Its Python submission contains real Python SyntaxError-triggering fragments and never runs — while its bug…
📊 Full Test Data
Rank 3 Read Report
92.0
/100
MiniMax-M3
Code 📅 2026-08
MiniMax-M3 scored 92.0/100 this batch. Its CSV parser has a genuine whitespace bug — not just a missed edge case — that…
📊 Full Test Data
Rank 2 Read Report
93.9
/100
Claude Sonnet 5
Code 📅 2026-08
Claude Sonnet 5 scored 93.9/100 this batch — a flawless bug-fix round and well-organized login form, with the same whitespace-before-quote edge case…
📊 Full Test Data
Rank 1 Read Report
61.4
/100
DeepSeek V4
Code 📅 2026-08
DeepSeek V4 scored 61.4/100 this batch. Its Python submission crashes with an AttributeError on almost every realistic input, and its JavaScript bug-fix…
📊 Full Test Data
Rank 5 Read Report
92.6
/100
Kimi 2.6
Code 📅 2026-08
Kimi 2.6 scored 92.6/100 this batch — a clean bug-fix round and functional login form, with the same whitespace-before-quote edge case that…
📊 Full Test Data
Rank 4 Read Report
92.8
/100
Doubao Seed 2.1
Code 📅 2026-08
Doubao Seed 2.1 scored 92.8/100 this batch, with the most thoroughly documented assumptions of the five models tested and a clean 7/7…
📊 Full Test Data
Rank 3 Read Report
93.2
/100
Qwen 3.7
Code 📅 2026-08
Qwen 3.7 scored 93.2/100 this batch — a clean 7/7 bug fix and solid frontend, with the same whitespace-before-quote edge case miss…
📊 Full Test Data
Rank 2 Read Report
95.2
/100
Grok 4.5
Code 📅 2026-08
Grok 4.5 led this month's panel with a 95.2/100, the only submission to pass all 8 Python tests including the whitespace-before-quote edge…
📊 Full Test Data
Rank 1 Read Report
92.0
/100
Gemini 3.1
Code 📅 2026-07
Gemini 3.1 is an AI coding model that scored 92.0/100 on this month's benchmark. It passed all 15 test cases across two…
📊 Full Test Data
Read Report
Explore More on AI Tool Scoop Reviews, comparisons, and everything you need to choose the right AI tools.
⭐ Reviews ⚔️ Compare 📋 Methodology
SUBMIT TOOL FREE