Public
AumeshMisra / tensorzero
Benchmark updated: 10/6/202630 Tasks
TensorZero is an open-source LLMOps platform that unifies an LLM gateway, observability, evaluation, optimization, and experimentation.
Languages
Rust79.5%TypeScript14.7%Python4%Shell0.6%Go0.4%PLpgSQL0.3%Other0.5%
Harness | Input / Output Cost | ||||||
|---|---|---|---|---|---|---|---|
1 |  Mini-SWE-agent | 26 / 30 | $0.58 | $4/$20 | 2m45s | ||
2 | Mini-SWE-agent | 26 / 30 | $1.39 | N/A | 4m00s | ||
3 | Mini-SWE-agent | 26 / 30 | $0.09 | $0.2/$1.2 | 4m58s |
Key Takeaways
- GPT-5.6 Sol (High) with Mini-SWE-agent is fastest at 165.32 seconds and costs $0.58 per test.
- GPT-5.6 Luna with Mini-SWE-agent costs least at $0.09 per test, while Claude Opus 5 (High Effort) costs $1.39.
Cost Analysis
Cost / Test vs. Accuracy
ACCURACYCOST
Average Token Use / Test
Token Usage
InputOutputReasoningCache readCache write
No token usage data available.
Cost is the clearest tradeoff in this comparison. anthropic/claude-opus-5-high leads at 86.67% for $1.39 per test. GPT-5.6 Luna is the lower-cost option at 86.67% for $0.09 per test.
Latency Analysis
Latency vs. Accuracy
ACCURACYLATENCY
Average Response Time / Test
Response Time
GPT-5.6 Luna
anthropic/claude-opus-5-high
GPT-5.6 Sol
Latency separates several models with similarly strong scores. anthropic/claude-opus-5-high leads at 86.67%, while GPT-5.6 Sol is fastest at 2m 45s with 86.67% accuracy.