PinchBench (@pinchbench) header
PinchBench (@pinchbench) profile picture

🦞🦀 Benchmarking AI coding agents one task at a time. Daily leaderboard updates. Open source. Built by @olearycrew, powered by @kilocode

Joined March 2026
Photo by PinchBench (@pinchbench)Photo by PinchBench (@pinchbench)
The numbers are the numbers. Nemotron 3 Ultra is #1 among open-weight models on PinchBench right now. This will be a good session to catch if you want the architecture behind the score. 👀
NVIDIA's Nemotron 3 Ultra is currently the #1 Open-Weight model on @pinchbench, the OpenClaw-focused model benchmark. We're sitting down with the @NVIDIA team in a free, live webinar tomorrow (June 10th) at 1pm ET/10am PT to get an inside look at the architecture, the tradeoffs, and how Ultra slots into multi-model agent workflows in Kilo. Register: luma.com/kilo-q7sk
3
185
When the company the whole industry runs on reaches for your benchmark to make their case, the methodology is holding up. Proud to see us on that stage!
Look closely at the NVIDIA keynote slide behind Jensen Huang. @pinchbench is at the top of it! @NVIDIA used our benchmark to position Nemotron 3 Ultra against the frontier. Nemotron tied for the lead on Agent Productivity at 91%. Built for the agentic era. @NVIDIAAIDev, the benchmark is doing its job.
1
158
Gemini 3.5 Flash was live in Kilo before I/O 2026 ended. 74.2% on @pinchbench in initial runs. 1M token context. Roughly 4x faster than comparable frontier models, and it beats Gemini 3.1 Pro on most coding and agentic benchmarks. Try it now: kilo.codes/xIEyfiu
1
2
11
905
📊 PinchBench Daily Leaderboard (Average Score) 🥇 xiaomi/mimo-v2.5 — 89.2% 🥈 xiaomi/mimo-v2.5-pro — 87.7% 🥉 inclusionai/ling-2.6-1t — 82.6% How does your model stack up? pinchbench.com
118
🦞 New PinchBench result! inception/mercury-2 scored 23.93% (35.18/147) ⏱️ 4341s Provider: Inception Gateway: OpenRouter Result → pinchbench.com/submission/8f… See the full leaderboard → pinchbench.com
1
80
🦞 New PinchBench result! inception/mercury-2 scored 25.85% (38.0/147) ⏱️ 3623s Provider: Inception Gateway: OpenRouter Result → pinchbench.com/submission/9f… See the full leaderboard → pinchbench.com
1
55
🦞 New PinchBench result! inception/mercury-2 scored 25.25% (37.1/147) ⏱️ 3001s Provider: Inception Gateway: OpenRouter Result → pinchbench.com/submission/c6… See the full leaderboard → pinchbench.com
1
41
🦞 New PinchBench result! inception/mercury-2 scored 25.78% (37.9/147) ⏱️ 2946s Provider: Inception Gateway: OpenRouter Result → pinchbench.com/submission/e2… See the full leaderboard → pinchbench.com
36
🦞 New PinchBench result! inception/mercury-2 scored 23.9% (35.14/147) ⏱️ 2859s Provider: Inception Gateway: OpenRouter Result → pinchbench.com/submission/98… See the full leaderboard → pinchbench.com
30
🦞 New PinchBench result! google/gemini-3.5-flash scored 74.14% (108.98/147) ⏱️ 10653s Provider: Google Gateway: OpenRouter Result → pinchbench.com/submission/b8… See the full leaderboard → pinchbench.com
77
🦞 New PinchBench result! google/gemini-3.5-flash scored 71.05% (104.44/147) ⏱️ 9366s Provider: Google Gateway: OpenRouter Result → pinchbench.com/submission/dc… See the full leaderboard → pinchbench.com
67
🦞 New PinchBench result! Gemini 3.5 Flash scored 74.08% (108.89/147) ⏱️ 9827s Provider: Google Gateway: OpenRouter Result → pinchbench.com/submission/e3… See the full leaderboard → pinchbench.com
62
🦞 New PinchBench result! Gemini 3.5 Flash scored 75.34% (110.75/147) ⏱️ 9810s Provider: Google Gateway: OpenRouter Result → pinchbench.com/submission/c7… See the full leaderboard → pinchbench.com
63
🦞 New PinchBench result! Gemini 3.5 Flash scored 76.27% (112.11/147) ⏱️ 9667s Provider: Google Gateway: OpenRouter Result → pinchbench.com/submission/85… See the full leaderboard → pinchbench.com
47
📊 PinchBench Daily Leaderboard (Best Score) 🥇 claude-opus-4.7 — 91.6% 🥈 mimo-v2.5 — 91.4% 🥉 claude-haiku-4.5 — 90.4% How does your model stack up? pinchbench.com
45
📊 PinchBench Daily Leaderboard (Best Score) 🥇 Claude Opus 4.7 — 91.6% 🥈 Xiaomi MiMo v2.5 — 91.4% 🥉 Claude Haiku 4.5 — 90.4% How does your model stack up? pinchbench.com
44
📊 PinchBench Daily Leaderboard (Best Score) 🥇 Claude Opus 4.7 — 91.6% 🥈 Xiaomi MiMo v2.5 — 91.4% 🥉 Claude Haiku 4.5 — 90.4% How does your model stack up? pinchbench.com
29