ministral-3:14b
Family: ministral-3 · 14B parameters · 1 run(s)
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): not measured (no valid ollama ps size for this run; tier kept as published). Tier rule
39.7/100 · scoring v4
95% task-bootstrap CI: 33.8–45.5 · ≈ statistically tied with other models in tie group 4 of this tier · rank #12 in tier, rank range under weight sensitivity #12–#12 (5th–95th percentile; min #11, max #16) · v3: 39.7 · mention 0.0% (0/45) · how
Plain-language summary
Weak as an agent today on this suite.
Good for
- —
Weak at
- matching the requested format
- 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 |
|---|---|---|---|---|---|---|---|
| 61 | 44 | 64 | 16 | 31 | — | 11.4 | 15.1 |
Prompt injection (15 tasks × 3 trials): useful and safe 31.1% (task completed and not compromised; counts in the score) · utility under attack —% (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: 31.1%. v2.1 used injection resistance 100.0 × tool-call rate . Definitions
- Run:
20261112-ministral-3-14b-7db599· 2026-11-12T09:36:18+00:00 - GPU: NVIDIA GeForce RTX 3090 (24 GB) · Linux 6.8.0-57-generic
- Host:
· kind not recorded · measured · 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)
- GGUF sha256:
bfb40fc6bb9c3b2ed529b480e04f824c005ea8f86733d4ebbf0c204de484891e - Sponsor disclosure: none
Chat template vs upstream
No reference
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.