GPT-5.6 Terra vs MiMo V2.6 Flash: Benchmark Comparison

GPT-5.6 Terra has the higher score on 8 of 14 shared benchmarks; MiMo V2.6 Flash leads on 5.

The largest observed score gap is 39.94 pts on IOI , where GPT-5.6 Terra leads.

Reported ±1 standard-error ranges overlap on 10 of 14 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 GPT-5.6 Terra MiMo V2.6 Flash Gap Reported uncertainty
Legal Research Bench 41.35% ±3.42 37.98% ±3.37 3.36 pts Reported ±1 SE ranges overlap
Tax Agent Bench 30.04% ±3.31 28.51% ±3.14 1.53 pts Reported ±1 SE ranges overlap
MedCode 43.41% ±2.17 41.06% ±2.03 2.36 pts Reported ±1 SE ranges overlap
MedScribe 82.87% ±1.95 85.28% ±1.99 2.41 pts Reported ±1 SE ranges overlap
ProofBench v1.1 74.00% ±4.41 63.00% ±4.85 11.00 pts Reported ±1 SE ranges do not overlap
Terminal-Bench-Science 10.00% ±3.61 4.29% ±2.44 5.71 pts Reported ±1 SE ranges overlap
SAGE 87.04% ±2.00 83.26% ±2.14 3.78 pts Reported ±1 SE ranges overlap
IOI 87.61% ±6.53 47.67% ±4.79 39.94 pts Reported ±1 SE ranges do not overlap
ProgramBench 0.50% ±0.50 0.50% ±0.50 0.00 pts Reported ±1 SE ranges overlap
Terminal-Bench 4.0 22.73% ±2.31 24.24% ±1.51 1.52 pts Reported ±1 SE ranges overlap
Vibe Code Bench v1.1 74.59% ±4.20 78.96% ±4.04 4.37 pts Reported ±1 SE ranges overlap
CyberBench v1.1 72.08% ±5.41 75.36% ±5.42 3.27 pts Reported ±1 SE ranges overlap
SRE Bench 9.54% ±1.82 4.20% ±1.24 5.34 pts Reported ±1 SE ranges do not overlap
Public Benefits Bench v1.1 62.38% ±1.26 67.59% ±1.22 5.21 pts Reported ±1 SE ranges do not overlap

Performance by category

Category GPT-5.6 Terra average MiMo V2.6 Flash average
Legal 41.35% 37.98%
Finance 30.04% 28.51%
Healthcare 63.14% 63.17%
Math 74.00% 63.00%
Science 10.00% 4.29%
Education 87.04% 83.26%
Coding 46.36% 37.84%
Cyber 40.81% 39.78%
Social Mobility 62.38% 67.59%

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 GPT-5.6 Terra cost MiMo V2.6 Flash cost GPT-5.6 Terra latency MiMo V2.6 Flash latency
Legal Research Bench $7.91 $0.07 1h10m 15m20s
Tax Agent Bench $4.83 $0.05 1h01m 17m19s
MedCode N/A N/A 18.41s 93.37s
MedScribe N/A N/A 35.53s 74.48s
ProofBench v1.1 $1.27 $0.12 12m57s 1h03m
Terminal-Bench-Science $5.19 $0.24 2h04m 3h59m
SAGE $0.11 $0.01 42.92s 2m47s
IOI $8.66 $0.22 1h23m 1h49m
ProgramBench $5.38 $0.47 33m36s 3h22m
Terminal-Bench 4.0 $5.60 $0.21 31m29s 2h25m
Vibe Code Bench v1.1 $7.89 $0.56 19m12s 50m13s
CyberBench v1.1 $3.31 $0.05 17m04s 28m22s
SRE Bench $30.86 $0.40 3h17m 3h22m
Public Benefits Bench v1.1 $1.20 $0.03 19m31s 20m56s

Results available only for GPT-5.6 Terra

  • LegalBench
  • MortgageTax
  • TaxEval v2
  • GPQA Diamond
  • MMLU Pro
  • MMMU Pro
  • LiveCodeBench
  • SkillsBench
  • SWE-bench
  • Financial Coaching Bench

Results available only for MiMo V2.6 Flash

  • Vals Index
  • Vals Finance Index
  • Vals Legal Index
  • Vals Science Index
  • HLAB-Vals-1.1
  • EMB v1.1
  • Finance Agent v2.1
  • BioMysteryBench
  • MysteryMechanism
  • Code Migration v1.1
  • Vibe Code Bench 1-100 v1.1
  • Access to Justice Benchmark
Model details GPT-5.6 Terra Model details MiMo V2.6 Flash