← Benches
gate.tools.result_use
GategauntletA deterministic gate probe — harness-invariant vitals / fidelity / tool-use check that every model must clear before the ladders count.
Solved
23/23
runs passed
Models
22
have attempted
Harnesses
2
scaffolds tried
Runs (latest per model × harness — solved sorted first)
| Model | Harness | Result | Turns | tok/s | Latency | Ctx | When |
|---|---|---|---|---|---|---|---|
| gpt-oss-120b | baseline | PASS | 2 | 43.6 | 1.4s | 64k | 2026-06-23 |
| granite-4.1-8b | baseline | PASS | 2 | 21.8 | 1.6s | 64k | 2026-06-23 |
| llama-3.3-70b | baseline | PASS | 2 | 2.7 | 12.1s | 64k | 2026-06-23 |
| devstral-small-2-2512 | baseline | PASS | 2 | 8.4 | 3.0s | 64k | 2026-06-23 |
| mistral-small-3.1-24b-instruct-2503 | baseline | PASS | 2 | 7.9 | 5.3s | 128k | 2026-06-23 |
| qwen3.5-122b-a10b-cluster | baseline | PASS | 2 | 15.0 | 7.1s | 64k | 2026-06-29 |
| qwen3.5-4b | baseline | PASS | 2 | 43.5 | 3.1s | 128k | 2026-06-30 |
| gpt-oss-20b | baseline | PASS | 2 | 67.3 | 1.2s | 64k | 2026-07-01 |
| qwen3-coder-30b | baseline | PASS | 2 | 29.3 | 1.2s | 64k | 2026-07-01 |
| gemma-4-26b-a4b-it | baseline | PASS | 2 | 22.5 | 1.4s | 64k | 2026-07-01 |
| qwen3.6-35b-a3b-or | baseline | PASS | 2 | 55.4 | 2.6s | 64k | 2026-07-01 |
| gemini-2.5-flash-lite | baseline | PASS | 2 | 2.1 | 8.7s | 64k | 2026-07-01 |
| qwen3.5-0.8b | baseline | PASS | 2 | 65.7 | 548ms | 64k | 2026-07-01 |
| qwen3.6-27b | baseline | PASS | 2 | 7.8 | 33.2s | 64k | 2026-07-01 |
| deepseek-r1-distill-qwen-32b | baseline | PASS | 2 | 4.7 | 1.4m | 64k | 2026-07-02 |
| qwen3.6-35b-bf16 | baseline | PASS | 2 | 32.2 | 7.2s | 64k | 2026-07-02 |
| qwen3.5-2b | baseline | PASS | 2 | 74.0 | 527ms | 64k | 2026-07-02 |
| granite-4.1-30b | baseline | PASS | 2 | 7.2 | 5.4s | 64k | 2026-07-02 |
| qwen3.6-27b | v1.7.1 | PASS | 2 | 7.6 | 5.4s | 64k | 2026-07-02 |
| qwen3.6-35b-a3b | baseline | PASS | 2 | 42.3 | 970ms | 64k | 2026-07-02 |
| qwen3-next-80b | baseline | PASS | 2 | 33.7 | 1.0s | 64k | 2026-07-02 |
| phi-4-reasoning-plus | baseline | PASS | 3 | 14.2 | 36.7s | 64k | 2026-07-02 |
| claude-sonnet-4.6 | baseline | PASS | 2 | 22.6 | 4.3s | 195k | 2026-07-06 |
Transcripts
gpt-oss-120b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The user with ID 7 is **Ada Lovelace**."}granite-4.1-8b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace."}llama-3.3-70b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The user with id 7 is named Ada Lovelace."}devstral-small-2-2512 · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace."}mistral-small-3.1-24b-instruct-2503 · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace."}qwen3.5-122b-a10b-cluster · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace."}qwen3.5-4b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "\n\nUser 7's name is Ada Lovelace."}gpt-oss-20b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "User 7 is **Ada Lovelace**."}qwen3-coder-30b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace.\n"}gemma-4-26b-a4b-it · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "<|channel>thought\n<channel|>The name of user 7 is Ada Lovelace."}qwen3.6-35b-a3b-or · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace."}gemini-2.5-flash-lite · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "Ada Lovelace is the name of user 7."}qwen3.5-0.8b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "User 7 is Ada Lovelace."}qwen3.6-27b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "\n\nUser 7's name is Ada Lovelace."}deepseek-r1-distill-qwen-32b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "\n\nThe user with ID 7 is named Ada Lovelace."}qwen3.6-35b-bf16 · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "\n\nUser 7's name is Ada Lovelace."}qwen3.5-2b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is Ada Lovelace."}granite-4.1-30b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The user with ID 7 is **Ada Lovelace**."}qwen3.6-27b · v1.7.1 · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "\n\nThe name of user 7 is Ada Lovelace."}qwen3.6-35b-a3b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "\n\nThe name of user 7 is Ada Lovelace."}qwen3-next-80b · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The user with ID 7 is named Ada Lovelace."}phi-4-reasoning-plus · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}, {"turn": 2, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "User says: \"Look up user 7 and tell me their name.\" The conversation includes a tool call result. We need to produce answer that uses the provided tool result.\n\nThe instructions say: \"Use the provided tool to look up data, then answer.\" So we have a tool result block with JSON content:\n[TOOL_RESULT]{\"user_id\": 7, \"name\": \"Ada Lovelace\", \"role\": \"engineer\"}[END_TOOL_RESULT]\n\nWe need to produce an answer that says: \"User 7's name is Ada Lovelace.\" But instructions say: \"Answer normally if no tools are needed.\"\n\nBut we have a tool result. The conversation includes the tool call and then the tool result block. We must use the provided tool result to inform our response.\n\nWe need to produce answer that says something like: \"User 7 is Ada Lovelace.\" But instructions say: \"Look up user 7 and tell me their name.\"\n\nI'll produce answer: \"Ada Lovelace\" or \"The name of user 7 is Ada Lovelace.\" I must not include any tool result tags in my answer. The instructions say: \"Never output [TOOL_RESULT] or [END_TOOL_RESULT] tags yourself.\" So I should simply produce a plain text answer.\n\nI'll produce answer: \"User 7's name is Ada Lovelace.\"\n\nI'll produce answer as final message. We'll produce answer with no markdown formatting, just plain text.\n\nI'll produce answer: \"The user with ID 7 is named Ada Lovelace.\" I'll produce answer in plain text.\n\nI'll produce answer: \"Ada Lovelace\" or \"User 7's name is Ada Lovelace.\"\n\nI'll produce answer as final message.User 7's name is Ada Lovelace."}claude-sonnet-4.6 · baseline · PASS
{"tool_calls": [{"turn": 1, "name": "get_user", "args": {"user_id": 7}, "args_ok": true}], "final": "The name of user 7 is **Ada Lovelace**! She is listed as an engineer in the system."}