Gemma3 1B IT (Heretic thinking)

model page · concurrency sweep 1–24 agents

Gemma3 1B IT (Heretic thinking)

server model id: gemma-3-1b-it-glm-4.7-flash-heretic-uncensored-thinking_gguf
thinkingKV F16ctx 32k1B params

Friendliest thinking profile — 86ms solo TTFT, 719 tok/s all-in, and it just thinks harder under load.

Key findings

  • Reasonable latency: TTFT stays in the 86ms–4.7s range across the whole sweep (single agent: 86ms) — dramatically better than minicpm5's 1.5s–52s. This is the friendliest "thinking" profile of the three models tested.
  • All-in throughput: counting reasoning, combined throughput reaches 719 tok/s at 24 agents (4.2x single-agent), mid-way between llama-3.2 (652) and minicpm (983). Visible content tops out around ~454 tok/s.
  • Thinking share grows with load: reasoning is only ~13% of tokens at concurrency 1, but climbs to ~45–58% at 17–24 — the model thinks proportionally more when slots contend.
  • Per-agent speed: falls from ~162–217 tok/s at low concurrency to ~36–77 at high concurrency — same slot-contention pattern as the other models.
  • Errors: 0 across all 24 runs.

Sweep — total concurrency 1–24

agentswall (s)contentreasontotalcomb tok/scomb allper-agentper-agent allscale %TTFT meanTTFT maxoktimeouterr
160.08,9221,35410,276149171162.2188.2100%86ms86ms010
260.014,3594,39318,752239312217.1284.4182%102ms114ms020
360.017,3705,82723,197289386118.4147.3226%1.4s1.4s030
460.020,4954,92225,41734142357.069.4247%2.3s2.3s040
560.019,7865,79225,57833042655.679.4249%2.7s2.7s050
660.018,6359,52628,16131046948.866.5274%2.5s2.5s060
760.019,7248,66928,39332947338.459.7277%3.2s3.2s070
860.019,5508,76328,31332647236.345.4276%4.1s4.2s080
960.026,8577,20734,06444756762.974.9332%3.7s3.7s090
1060.021,78714,63236,41936360758.681.6355%4.4s4.4s190
1160.023,55714,51538,07239263448.278.0371%2.9s2.9s0110
1260.024,20014,34338,54340364254.573.0375%3.5s3.5s1110
1360.024,93715,48940,42641567355.176.7394%3.8s3.8s1120
1460.027,27212,93040,20245467047.570.2392%4.1s4.1s0140
1560.024,44416,03840,48240767469.969.5394%4.3s4.3s4110
1660.024,86816,17641,04441468461.063.5400%3.6s3.6s4120
1760.022,07019,93342,00336870051.265.1409%4.0s4.0s4130
1860.024,55717,00741,56440969276.864.0405%4.7s4.7s7110
1960.025,39116,36841,75942369573.262.0406%4.1s4.1s8110
2060.023,77617,54941,32539668872.359.5402%4.2s4.3s10100
2160.018,10820,08938,19730263676.951.3372%4.6s4.6s1560
2260.022,24719,91442,16137070254.651.7411%2.5s2.5s1390
2360.016,25222,70338,95527164957.347.8380%3.6s3.6s1760
24 peak60.122,95820,21743,17538271955.255.4420%2.9s3.0s1590

* 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.
Visible content vs hidden reasoning tokens per run, plus the reasoning share (%).
Visible content vs hidden reasoning tokens per run, plus the reasoning share (%).

Raw data & downloads