mistral:7b
Family: mistral · 7B parameters · 2 run(s)
Why this model scores about 0 on tool selection: what it wrote instead of tool calls, and whether the template, the engine or the model is the cause. Read the analysis. Applies to all runs on this page; scores unchanged.
Fits in 16 GB VRAM · ollama 0.40.6 · Q4_K_M
Measured VRAM at 32K context (KV cache included, ollama ps on the RTX 3090): 8.9 GB · previous tier: 8 GB, rank #11 (results v0.1.1). Moved to the 16 GB tier in results v0.1.2 (2026-10-26) because it needs more than 8 GB at 32K context; score unchanged. Tier rule
22.0/100 · scoring v4
95% task-bootstrap CI (main suite, 3 trials): 19.8–24.2 · pooled over 8 runs (3 main-suite trials + 5 reruns at temperature 0.7): 21.2–25.2 · rank #16 in tier, rank range under weight sensitivity #16–#16 (5th–95th percentile; min #15, max #16) · v3: 22.0 (#14) · mention 0.0% (0/45) · how
Plain-language summary
Doesn't work as a tool-using agent in this suite.
Good for
- returning clean JSON
Weak at
- picking the right tool
- multi-step plans
- staying safe under prompt injection
Category breakdown
| JSON | Schema | Tool | Multi-step | Injection (useful&safe) | tok/s | p50 s/task | p95 s/task |
|---|---|---|---|---|---|---|---|
| 93 | 53 | 0 | 0 | 0 | 128.3 @37becd | 14.7 | 17.7 |
Prompt injection (15 tasks × 3 trials): useful and safe 0.0% (task completed and not compromised; counts in the score) · utility under attack 0.0% (completed the user's task) · harmful-action attack success 0.0% (a real harmful tool call or canary leak). Mentions (quoting the attacker without acting): 0.0% (0/45); they do not lower the score. v3 useful and safe: 0.0%. v2.1 used injection resistance 100.0 × tool-call rate 0. Definitions
- Run:
20261021-mistral-7b-83c59b· 2026-10-21T18:07:31+00:00 - GPU: NVIDIA GeForce RTX 3090 (24 GB) · Linux 6.8.0-49-generic
- Host:
37becd33558a· container (RunPod) · host AMD EPYC 7402P 24-Core Processor, 251.6 GB RAM · measured 2026-10-21T18:07:31+00:00 · hosts - Harness: hermes-agent Hermes Agent v0.22.2+2363.g9cb34e9 (2026.9.24) · stable · local 9cb34e9e285d · trials 3 · sampling: Ollama default (temperature and seed not set by harness) · context window: 32768 tokens as loaded by Ollama (the unused runner field records 8192; see Methodology)
- Score per trial (v4): 22.2 / 24.6 / 19.2 (SD 2.71, range 19.2–24.6)
- GGUF sha256:
f5074b1221da0f5a2910d33b642efa5b9eb58cfdddca1c79e16d7ad28aa2b31f - Sponsor disclosure: none
Real 8 GB card: NVIDIA RTX 3070 8 GB (RunPod container) · ollama 0.40.6 · Q4_K_M
Separate series, not part of the tier ranking. 3 single-trial run(s) merged, scoring v4 from the run logs. CPU/GPU split (ollama ps): 8.6 GB 19%/81% CPU/GPU runs with CPU offload (slow) (main tier stays 16 GB) · peak VRAM (nvidia-smi): 6813 MiB of 8192 MiB · series table · how it was measured
21.8/100 · scoring v4
95% task-bootstrap CI: 19.6–23.8 · rank #11 in this series, rank range under weight sensitivity #11–#11 · how
Plain-language summary
Doesn't work as a tool-using agent in this suite.
Good for
- returning clean JSON
Weak at
- picking the right tool
- multi-step plans
- staying safe under prompt injection
Category breakdown
| JSON | Schema | Tool | Multi-step | Injection (useful&safe) | tok/s | p50 s/task | p95 s/task |
|---|---|---|---|---|---|---|---|
| 91 | 55 | 0 | 0 | 0 | 21.5 @8fcb65 | — | — |
Prompt injection (15 tasks × 3 trials): useful and safe 0.0% (task completed and not compromised; counts in the score) · utility under attack 0.0% (completed the user's task) · harmful-action attack success 0.0% (a real harmful tool call or canary leak). Mentions (quoting the attacker without acting): 0.0% (0/45); they do not lower the score. v3 useful and safe: 0.0%. Definitions
- Run:
real8gb-mistral-7b· 2026-10-27T08:37:17+00:00 - Hardware: Real 8 GB card: NVIDIA RTX 3070 8 GB (RunPod container) · NVIDIA GeForce RTX 3070 · Linux 6.8.0-85-generic
- Host:
8fcb655dfb7d· container (RunPod) · pod b3o5tul56pnqfv · pod allotment 8 vCPU / 30 GB · host AMD Ryzen 7 5700X 8-Core Processor, 62.7 GB RAM · measured 2026-10-27T08:37:17+00:00 · hosts - Harness: hermes-agent Hermes Agent v0.22.2+2363.g9cb34e9 (2026.9.24) · stable · local 9cb34e9e285d · trials 3 · sampling: Ollama default (temperature and seed not set by harness) · context window: 32768 tokens as loaded by Ollama (the unused runner field records ; see Methodology)
- Score per trial (v4): 19.8 / 22.2 / 23.4 (SD 1.83, range 19.8–23.4)
- GGUF sha256:
f5074b1221da0f5a2910d33b642efa5b9eb58cfdddca1c79e16d7ad28aa2b31f - Sponsor disclosure: none
Chat template vs upstream
Differs: the GGUF template is not identical to the provider's current upstream template after whitespace normalisation. This is not a safety verdict: older revisions, Ollama conversions and legitimate re-uploads with fixed templates (e.g. Unsloth) often differ.
- GGUF template sha256 (normalised):
26a59556925c987317ce5291811ba3b7f32ec4c647c400c6cc7e3a9993007ba7 - Upstream reference sha256 (normalised):
603a4c4659c7bb1d0a2d666e5d4560b4b533dbbf6a3eb1d041509652cf7890f1· source
Labels are only “matches / differs / no reference”, never safe/unsafe. Ollama may use a built-in Go template or renderer instead of the GGUF Jinja. Why templates matter: Chat-template backdoors. Not part of the score.