Matrix overall
56.6
AVG 32.4 across matrix cells
Active
Model portfolio
StarCoder 2 15B
Tap a star to rate
BigCode · Closed · Context 16K tokens
Cadle Score
104.7
#56 in index
Benchmark average
32.4
Not enough benchmark results yet
Pricing
Unlisted
no list price
MODEL MATRIX · TRUTH CENTER CELLS
Overall
56.6
Intelligence
56.6
Coding
38.4
Reasoning
3.1
Language
27.8
Knowledge
33.5
Agentic
—
MMLU
33.5
GPQA
3.1
Math
57.7
LiveCodeBench
—
HumanEval
—
Reason
29.6
ARC-AGI
—
Coverage AVG
32.4
Pillar Coverage:6 / 7 Core Pillars (86%)
Cadle Cap ranks models with within-benchmark percentiles so ARC-AGI ~40 and MMLU ~90 never compete raw. Coverage AVG is an importance-weighted mean of filled expert cells only — blanks are skipped, not zeroed.
ANALYSIS · TRUTH CENTER
Hard benches (GPQA · ARC-AGI · LCB · Agentic) sit left — a mid-40 on ARC-AGI can still be frontier. Broader benches on the right are easier and often higher.
CADLE INSIGHTS
Only measured signals
Benchmark avg
31.5
13 tracked scores
Strongest signal
epoch-eci
Score 104.7
Blended $/1M
Unlisted
No published list price
Context
16K
Provider context window (not max output)
Coding (matrix)
38.4
Model Matrix coding cell
Intel index
56.6
Matrix AA IQ cell
SPEED · COST · QUALITY
Measured matrix and verified gateway observations only.
- Cadle Score
- 104.7
- Overall (Matrix)
- 56.6
- Cadle Intelligence
- 56.6
- Matrix Coding
- 38.4
DESCRIPTION + DETAILS
PROVIDER
BigCode
CONTEXT
16K tokens
STATUS
ACTIVE
COMMUNITY REVIEWS
Community Reviews & Field Notes
Verified EngineersReal operator evaluations, prompt robustness observations, and deployment notes for StarCoder 2 15B.
4.5(2)
Leave a Field Observation
Rating:5 - Frontier benchmark
Role:
Tag topic:
Reviews are publicly visible to the engineering community.
Inference OperatorProduction Deployer Verified
Nov 2, 2026
Tested StarCoder 2 15B on high-concurrency tool loops. Token generation remains consistent under sustained context load, with minimal prompt drift on complex JSON schemas.
Autonomous Agent LeadEval Maintainer Verified
Oct 29, 2026
Evaluated StarCoder 2 15B on multi-turn code synthesis. Code reasoning and diff application succeed on ~88% of test suites. Best paired with strict system prompt guardrails.
MODEL COMPARISON
| Metric |  StarCoder 2 15B |
|---|---|
| Provider | BigCode |
| Release | |
| Input / 1M | No data yet |
| Output / 1M | No data yet |
| Context | 16K tokens |
| Cadle Cap | 105 |
| Cap board | Cadle Language Cap 105 |
| bigcodebench | 38.4 |
| mmlu-pro | 15.0 |
| musr | 2.9 |
| gpqa-diamond | 3.1 |
| math-level-5 | 6.0 |
| bbh | 20.4 |
| ifeval | 27.8 |
| training-compute-flops | 23.6 |
| winogrande | 28.6 |
| mmlu | 52.1 |
| gsm8k | 57.7 |
| arc-ai2 | 29.6 |
DATA PROVENANCE
Scores may use different scales — check the Benchmarks tab for source links.
- Supabase models
- Cadle public observations
MEASURED CELLS
Same Truth Center numbers as the datasheet above. Accuracy, reliability, safety, and cost scores are not invented.
Overall56.6
AA IQ56.6
Coding38.4
MMLU33.5
GPQA3.1
Math57.7
TECHNICAL FIELD GUIDESAll 10 Guides →