MiniCPM5 1B ×2 (parallel)

model page · concurrency sweep 2–48 agents · 2 parallel instances

MiniCPM5 1B ×2 (parallel)

server model id: minicpm5-1b-claude-opus-fable5-v2-thinking / -2
2× parallel instancesthinkingKV F16ctx 32k1B params

The thinking champion doubled: two parallel LM Studio instances — 843 tok/s inside the 24-agent window, up to 1,698 tok/s at 48 total agents (6.4x).

Key findings

  • Two instances roughly double throughput again: combined all-token output reaches 1698 tok/s at 48 total agents (6.4x the 2-agent baseline) — vs the single-instance minicpm peak of ~983 tok/s. Visible content throughput also tops out at 835 tok/s at 48, well above the single-instance ~537.
  • Late-game jump: throughput holds ~830–1000 between 20–38 agents, then jumps to 1321–1698 at 40–48 — the two slots' batching kicks in hard at the top end.
  • Reasoning share climbs with load: ~63% of tokens at total-2, settling to ~25–55% in the mid-range, then back up to ~48–52% at 44–48.
  • TTFT is poor (thinking model): 2.4s at total-2, mostly 14–33s under load with max values up to 59.9s. Latency-critical use remains a bad fit.
  • Errors: 0 across all 24 levels.

Sweep — total concurrency 2–48

agentswall (s)contentreasontotalcomb tok/scomb allper-agentper-agent allscale %TTFT meanTTFT maxoktimeouterr
260.05,95710,00315,96099266117.9311.0100%2.4s2.9s020
460.018,64010,84929,489310491148.9180.6185%13.7s40.8s040
660.018,74216,10934,851312580114.5175.4218%18.3s45.0s060
860.019,36719,02238,38932363969.4114.0240%15.5s53.0s080
1060.027,46212,39639,85845766472.5103.9250%9.4s18.1s0100
1260.026,05018,73144,78143474680.6108.9280%12.3s59.9s0120
1460.026,61117,20743,81844373067.294.7274%21.2s43.5s0140
1660.030,26211,76842,03050470066.279.3263%23.4s33.0s0160
1860.030,66216,50647,16851178665.979.5295%24.3s54.7s0180
2060.036,14414,20050,34460283860.174.4315%23.3s32.2s0200
2260.035,37315,24750,62058984357.670.5317%26.9s38.2s0220
2460.034,02415,86149,88556783153.366.3312%29.5s41.4s0240
2660.038,18715,88054,06763690054.364.4338%27.8s45.9s0260
2860.037,12616,45853,58461889254.064.8335%31.9s52.2s0280
3060.042,30817,07459,38270598951.060.5372%30.0s44.4s0300
3260.034,69922,61057,30957895442.052.4359%32.6s57.9s0320
3460.140,07419,24759,32166798843.653.9371%32.0s39.5s0340
3660.043,90716,40360,3107311,00456.858.9377%32.9s48.3s0360
3860.037,91720,79158,70863197840.250.0368%33.4s45.8s0380
4060.048,22345,18993,4128031,55643.164.8585%17.4s56.9s0400
4260.043,99235,33279,3247331,32140.658.4497%22.3s57.3s0420
4460.149,30652,456101,7628211,69534.454.5637%14.6s48.6s0440
4660.041,36546,83688,2016891,46929.848.1552%22.7s52.6s0460
48 peak60.150,14451,793101,9378351,69838.555.0638%14.3s50.6s0480

Parallel run: N agents ran on each of two LM Studio instances simultaneously (total 2×N). * 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