Ember-1 vs Kimi K3: Benchmark Comparison
Ember-1 has the higher score on 2 of 10 shared benchmarks; Kimi K3 leads on 8.
The largest observed score gap is 26.00 pts on ProofBench v1.1 , where Kimi K3 leads.
Reported ±1 standard-error ranges overlap on 7 of 10 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 | Ember-1 | Kimi K3 | Gap | Reported uncertainty |
|---|---|---|---|---|
| Legal Research Bench | 44.71% ±3.46 | 46.15% ±3.46 | 1.44 pts | Reported ±1 SE ranges overlap |
| Tax Agent Bench | 31.08% ±3.34 | 33.37% ±3.39 | 2.29 pts | Reported ±1 SE ranges overlap |
| ProofBench v1.1 | 61.00% ±4.90 | 87.00% ±3.38 | 26.00 pts | Reported ±1 SE ranges do not overlap |
| BioMysteryBench | 67.04% ±3.03 | 72.59% ±1.61 | 5.56 pts | Reported ±1 SE ranges do not overlap |
| Terminal-Bench-Science | 5.71% ±2.79 | 1.43% ±1.43 | 4.29 pts | Reported ±1 SE ranges do not overlap |
| IOI | 46.33% ±9.16 | 48.94% ±9.82 | 2.61 pts | Reported ±1 SE ranges overlap |
| Terminal-Bench 4.0 | 20.71% ±3.64 | 17.17% ±0.51 | 3.54 pts | Reported ±1 SE ranges overlap |
| Vibe Code Bench v1.1 | 83.87% ±3.15 | 84.97% ±2.75 | 1.10 pts | Reported ±1 SE ranges overlap |
| CyberBench v1.1 | 69.52% ±5.56 | 75.24% ±5.56 | 5.71 pts | Reported ±1 SE ranges overlap |
| Public Benefits Bench v1.1 | 66.17% ±1.23 | 68.20% ±1.21 | 2.03 pts | Reported ±1 SE ranges overlap |
Performance by category
| Category | Ember-1 average | Kimi K3 average |
|---|---|---|
| Legal | 44.71% | 46.15% |
| Finance | 31.08% | 33.37% |
| Math | 61.00% | 87.00% |
| Science | 36.38% | 37.01% |
| Coding | 50.30% | 50.36% |
| Cyber | 69.52% | 75.24% |
| Social Mobility | 66.17% | 68.20% |
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 | Ember-1 cost | Kimi K3 cost | Ember-1 latency | Kimi K3 latency |
|---|---|---|---|---|
| Legal Research Bench | $3.67 | $3.47 | 38m34s | 12m40s |
| Tax Agent Bench | $2.97 | $2.79 | 30m46s | 40m15s |
| ProofBench v1.1 | $1.74 | $1.65 | 21m09s | 34m31s |
| BioMysteryBench | $0.91 | $1.06 | 9m26s | 4m55s |
| Terminal-Bench-Science | $30.76 | $21.30 | 3h23m | 4h49m |
| IOI | $24.46 | $14.17 | 1h47m | 4h18m |
| Terminal-Bench 4.0 | $19.48 | $12.48 | 1h55m | 3h09m |
| Vibe Code Bench v1.1 | $8.56 | $10.01 | 39m18s | 16m39s |
| CyberBench v1.1 | $2.43 | $2.13 | 22m02s | 30m44s |
| Public Benefits Bench v1.1 | $1.02 | $0.91 | 13m21s | 7m19s |
Results available only for Ember-1
- Vals Science Index
- MysteryMechanism
Results available only for Kimi K3
- Vals Finance Index
- Vals Index
- Vals Inference Provider Index
- Vals Legal Index
- Vals RSI Index
- HLAB-Vals-1.1
- LegalBench
- EMB v1.1
- Finance Agent v2.1
- MortgageTax
- TaxEval v2
- MedCode
- MedScribe
- GPQA Diamond
- MMLU Pro
- MMMU Pro
- SAGE
- Code Migration v1.1
- LiveCodeBench
- ProgramBench
- SWE-bench
- SWE-bench Multimodal
- Vibe Code Bench 1-100
- Time Horizon Index: KSP