Kimi K2.6 vs GLM 5.3 Flash: Benchmark Comparison
Kimi K2.6 has the higher score on 8 of 16 shared benchmarks; GLM 5.3 Flash leads on 7.
The largest observed score gap is 29.33 pts on Legal Research Bench , where GLM 5.3 Flash leads.
Reported ±1 standard-error ranges overlap on 8 of 16 shared benchmarks with comparable uncertainty data. This is not a pairwise statistical significance test.
Shared benchmark results
Scores and reported standard errors come from the latest visible Vals benchmark version available for both models.
| Benchmark | Kimi K2.6 | GLM 5.3 Flash | Gap | Reported uncertainty |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | 1.67% ±0.83 | 6.67% ±1.65 | 5.00 pts | Reported ±1 SE ranges do not overlap |
| Legal Research Bench | 15.87% ±2.54 | 45.19% ±3.46 | 29.33 pts | Reported ±1 SE ranges do not overlap |
| LegalBench | 84.74% ±0.45 | 83.93% ±0.43 | 0.81 pts | Reported ±1 SE ranges overlap |
| EMB | 57.85% ±2.74 | 55.93% ±3.56 | 1.92 pts | Reported ±1 SE ranges overlap |
| Finance Agent (v2) | 44.90% ±0.73 | 57.85% ±2.06 | 12.95 pts | Reported ±1 SE ranges do not overlap |
| Tax Agent Bench | 45.13% ±3.23 | 61.96% ±3.26 | 16.82 pts | Reported ±1 SE ranges do not overlap |
| TaxEval v2 | 74.65% ±0.85 | 75.59% ±0.84 | 0.94 pts | Reported ±1 SE ranges overlap |
| MedScribe | 78.15% ±1.79 | 88.94% ±1.91 | 10.79 pts | Reported ±1 SE ranges do not overlap |
| GPQA Diamond | 89.14% ±2.00 | 86.36% ±2.13 | 2.78 pts | Reported ±1 SE ranges overlap |
| MMLU Pro | 87.57% ±0.33 | 86.06% ±0.34 | 1.51 pts | Reported ±1 SE ranges do not overlap |
| MMMU Pro | 86.30% ±0.83 | 86.01% ±0.83 | 0.29 pts | Reported ±1 SE ranges overlap |
| Code Migration | 27.77% ±4.14 | 20.52% ±4.21 | 7.25 pts | Reported ±1 SE ranges overlap |
| LiveCodeBench | 86.77% ±0.97 | 80.51% ±1.05 | 6.26 pts | Reported ±1 SE ranges do not overlap |
| ProgramBench | 0.00% ±0.00 | 0.00% ±0.00 | 0.00 pts | Reported ±1 SE ranges overlap |
| SWE-bench | 76.20% ±1.91 | 92.00% ±1.21 | 15.80 pts | Reported ±1 SE ranges do not overlap |
| Vibe Code Bench v1.1 | 37.89% ±4.91 | 30.76% ±5.28 | 7.13 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Kimi K2.6 average | GLM 5.3 Flash average |
|---|---|---|
| Legal | 34.09% | 45.26% |
| Finance | 55.64% | 62.83% |
| Healthcare | 78.15% | 88.94% |
| Academic | 87.67% | 86.14% |
| Coding | 45.73% | 44.76% |
Cost and latency
Cost per test appears only where the benchmark reports it for both models. Latency is the measured completion time for that benchmark.
| Benchmark | Kimi K2.6 cost | GLM 5.3 Flash cost | Kimi K2.6 latency | GLM 5.3 Flash latency |
|---|---|---|---|---|
| Harvey's Legal Agent Benchmark | $0.82 | $0.57 | 23m23s | 52m17s |
| Legal Research Bench | $1.52 | $0.12 | 30m53s | 51m07s |
| LegalBench | N/A | N/A | 50.07s | 12.52s |
| EMB | $2.20 | $0.71 | 39m48s | 1h15m |
| Finance Agent (v2) | $1.14 | $0.05 | 27m10s | 14m51s |
| Tax Agent Bench | $0.57 | $0.14 | 15m24s | 21m37s |
| TaxEval v2 | N/A | N/A | 4m27s | 53.06s |
| MedScribe | N/A | N/A | 8m14s | 86.98s |
| GPQA Diamond | N/A | N/A | 9m05s | 2m11s |
| MMLU Pro | N/A | N/A | 2m36s | 44.76s |
| MMMU Pro | N/A | N/A | 4m02s | 28.22s |
| Code Migration | $5.11 | $3.13 | 1h30m | 3h26m |
| LiveCodeBench | N/A | N/A | 8m49s | 5m02s |
| ProgramBench | N/A | N/A | 1h20m | 5h30m |
| SWE-bench | $0.27 | $0.02 | 8m40s | 1h04m |
| Vibe Code Bench v1.1 | $1.93 | $2.35 | 49m28s | 1h24m |
Results available only for Kimi K2.6
- MortgageTax
- MedCode
- SAGE
- Public Benefits Bench v1.1
Results available only for GLM 5.3 Flash
- ProofBench v1.1
- Terminal-Bench Science
- IOI
- SkillsBench
- Vibe Code Bench 1-100
- CyberBench v1.1