Claude Sonnet 4 vs GLM-4.6
Wins 3 of 6 areas
Coding · Long documents · Following instructions
Wins 2 of 6 areas
Agents · Reasoning
Claude Sonnet 4 wins more areas, narrowly.GLM-4.6 is cheaper and better at agents.
Scores updated · 42 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software52Claude Sonnet 45 of 7 tests
- Following instructionsDoing exactly what it is asked20Claude Sonnet 42 of 2 tests
- Long documentsFinding answers in very long texts10Claude Sonnet 41 of 1 test
- AgentsCarrying out multi-step tasks on its own12GLM-4.62 of 3 tests
- ReasoningHard problems that need careful thinking01GLM-4.61 of 3 tests · 2 ties
- FactsGetting facts right instead of making them up11Even1 each · 1 tie
Long documents rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
GLM-4.6 costs 85% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Sonnet 4 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+65points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+33points ahead
- Reasons across sets of long documentsAA-LCR+16.3points ahead
Where GLM-4.6 pulls ahead
- Recent programming contest problemsLiveCodeBench v6+34.3points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+32.9points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+13.1points ahead
Every test, side by side
All 42 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 4
- LiveCodeBench v6GLM-4.6 by 34.348.582.8+34.3
- SWE-bench ProClaude Sonnet 4 by 3342.79.7+33
- Terminal-Bench 2.1GLM-4.6 by 13.136.349.4+13.1
- Terminal-Bench HardClaude Sonnet 4 by 6.131.125+6.1
- SWE-bench VerifiedClaude Sonnet 4 by 4.772.768+4.7
- SWE-bench MultilingualClaude Sonnet 4 by 3.156.953.8+3.1
- SciCodeClaude Sonnet 4 by 1.64038.4+1.6
AgentsGLM-4.6
- BrowseCompGLM-4.6 by 32.912.245.1+32.9
- GDPValGLM-4.6 by 3.39.713+3.3
- τ-Bench V3 · BankingClaude Sonnet 4 by 3.316.713.4+3.3
ReasoningGLM-4.6
- Humanity's Last ExamGLM-4.6 by 3.810.714.5+3.8
- CritPttie0.31.1tie
- GPQA Diamondtie77.778tie
FactsEven
- AA-Omniscience · Non-hallucinationClaude Sonnet 4 by 6570.95.9+65
- AA-Omniscience · AccuracyGLM-4.6 by 4.222.726.9+4.2
- Vectara HHEM hallucination ratelower is bettertie10.39.5tie
Following instructionsClaude Sonnet 4
- IFBenchClaude Sonnet 4 by 11.354.743.4+11.3
- Multi-ChallengeClaude Sonnet 4 by 2.257.154.9+2.2
Other results23 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- HMMT 2025GLM-4.6 by 72.815.988.7+72.8
- AA-OmniscienceClaude Sonnet 4 by 42.10.2-41.9+42.1
- τ²-BenchGLM-4.6 by 3045.275.2+30
- AIME 2025GLM-4.6 by 23.470.593.9+23.4
- BrowseComp-ZHGLM-4.6 by 20.429.149.5+20.4
- FinSearchComp-globalClaude Sonnet 4 by 12.84229.2+12.8
- LiveCodeBenchGLM-4.6 by 1168.579.5+11
- HLE (with tools)GLM-4.6 by 10.120.330.4+10.1
- Artificial Analysis Coding IndexGLM-4.6 by 8.237.645.8+8.2
- vectara_avg_summary_lengthClaude Sonnet 4 by 68.6 rating points145.877.2+68.6 rating
- τ²-Bench Telecom (AA run)GLM-4.6 by 5.964.670.5+5.9
- Multi-SWE-BenchClaude Sonnet 4 by 5.735.730+5.7
- xbench-DeepSearchGLM-4.6 by 5.464.670+5.4
- Terminal-BenchGLM-4.6 by 535.540.5+5
- vectara_answer_rateClaude Sonnet 4 by 4.198.694.5+4.1
- GAIA (text only)GLM-4.6 by 3.668.371.9+3.6
- ArtifactsBenchGLM-4.6 by 2.557.359.8+2.5
- frontiermath_tier_4_v1GLM-4.6 by 2.102.1+2.1
- theagentcompanyClaude Sonnet 4 by 23735+2
- AA Agentic IndexGLM-4.6 by 117.618.6+1
- vectara_factual_consistencytie89.790.5tie
- MMLU-Protie83.783.2tie
- AA Intelligencetie18.918.5tie
Questions people ask
Which is better, Claude Sonnet 4 or GLM-4.6?
Claude Sonnet 4 wins three of the six areas where both have results: coding, long documents and following instructions. GLM-4.6 wins agents and reasoning, and costs 85% less. They are level on facts.
Which is better for coding?
Claude Sonnet 4. It wins 5 of the 7 coding tests both models report; GLM-4.6 wins 2.
Which is cheaper?
Claude Sonnet 4 costs $3.00 per million input tokens and $15.00 per million output tokens; GLM-4.6 costs $0.57 and $2.20. That makes GLM-4.6 about 85% cheaper for the same work.
How do you compare the two?
We use the 42 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 23 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.