Unsloth Ministral 3 3B (2512)

model page · concurrency sweep 1–24 agents

Unsloth Ministral 3 3B (2512)

server model id: unsloth/ministral-3-3b-instruct-2512
non-thinkingKV Q4_0ctx 32k3B params

Slowest solo of any model tested (47 tok/s on Q4_0 KV) but the highest relative scaling on record — 8.7x to 410 tok/s at 23 agents.

Key findings

  • Clean, all-content output: 0 reasoning tokens everywhere.
  • Slowest single-agent of every model tested: only 47 tok/s at 1 agent — notably slower than granite-4.1-3b (79), chronos (88), and llama-3.2-3b (95). With the Q4_0 KV quant, the KV cache path is 4-bit; on this setup that quant appears to hurt rather than help single-stream speed.
  • Giant relative scaling (highest recorded): combined climbs to 410 tok/s at 23 agents (8.7x) — the largest multiplier of any run — but only because the solo baseline is so tiny. Absolute peak (~375–410 tok/s from 13–24) is still mid-pack vs other models.
  • Noisy mid-range: big swings between 5–8 agents (106–184) and the 9–13 transition (297 → 253 → 365), showing erratic server batching behavior for this model.
  • TTFT: 112ms solo → 17.7s at 24 agents; a ~26.5s outlier at concurrency 12.
  • Per-agent speed: ~24–34 tok/s at high concurrency (vs 65 solo).
  • Errors: 0 across all 24 runs.

Sweep — total concurrency 1–24

agentswall (s)contentreasontotalcomb tok/scomb allper-agentper-agent allscale %TTFT meanTTFT maxoktimeouterr
160.02,83502,835474765.565.5100%115ms115ms010
260.04,79904,799808040.740.7170%112ms127ms020
360.010,015010,01516716788.588.5355%2.6s2.6s030
460.011,075011,07518418454.154.1391%5.4s5.4s040
560.06,38706,38710610623.623.6226%5.8s5.8s050
660.07,19007,19012012021.221.2255%3.5s3.5s060
760.08,40708,40714014021.421.4298%3.8s3.8s070
860.08,12708,12713513518.718.7287%5.7s5.7s080
960.017,849017,84929729737.137.1632%5.3s5.4s090
1060.017,841017,84129729738.038.0632%11.4s11.4s0100
1160.018,324018,32430530536.136.1649%11.1s11.1s0110
1260.015,193015,19325325337.837.8538%26.5s26.5s0120
1360.021,925021,92536536533.533.5777%9.5s9.6s0130
1460.021,584021,58435935933.933.9764%12.9s12.9s0140
1560.022,363022,36337237231.831.8791%12.9s12.9s0150
1660.022,537022,53737537530.830.8798%14.2s14.3s0160
1760.022,218022,21837037027.827.8787%13.0s13.1s0170
1860.022,455022,45537437426.826.8796%13.4s13.5s0180
1960.022,909022,90938238225.925.9813%13.3s13.5s0190
2060.022,237022,23737037025.525.5787%16.5s16.6s0200
2160.023,800023,80039639624.724.7843%14.1s14.3s0210
2260.023,630023,63039439424.324.3838%15.7s15.9s0220
23 peak60.024,605024,60541041023.823.8872%15.0s15.2s0230
2460.024,005024,00540040023.623.6851%17.7s17.9s0240

* burst artifact — agents queued ~11–16s then generated in a short burst; the per-agent timer inflates the number. Trust combined throughput and TTFT columns. Peak row = highest combined (all-token) throughput.

Charts

Whole sweep at a glance — combined throughput, scaling efficiency, per-agent throughput and outcomes across every concurrency level.
Whole sweep at a glance — combined throughput, scaling efficiency, per-agent throughput and outcomes across every concurrency level.
Combined throughput (all tokens) vs concurrent agents. Annotated peak with the concurrency level where it occurred.
Combined throughput (all tokens) vs concurrent agents. Annotated peak with the concurrency level where it occurred.
Per-agent throughput erosion as slots contend. Red stars mark burst-timing artifacts (queued ≥ 10s then generated in a burst — inflated values, not real throughput).
Per-agent throughput erosion as slots contend. Red stars mark burst-timing artifacts (queued ≥ 10s then generated in a burst — inflated values, not real throughput).
Scaling efficiency relative to the single-agent baseline (100%).
Scaling efficiency relative to the single-agent baseline (100%).
Time to first token — mean (solid) and max (dashed). Log scale. Gaps mean no visible content ever arrived within the 60s window.
Time to first token — mean (solid) and max (dashed). Log scale. Gaps mean no visible content ever arrived within the 60s window.
Total tokens per 60s run — all tokens vs visible content only.
Total tokens per 60s run — all tokens vs visible content only.
How each run of N agents finished: completed, timed out (60s budget), or errored.
How each run of N agents finished: completed, timed out (60s budget), or errored.
The concurrency trade-off on one chart — combined throughput rises while per-agent speed falls.
The concurrency trade-off on one chart — combined throughput rises while per-agent speed falls.

Raw data & downloads