This is the short archive entry for the June DeepSeek V4 Flash work. The detailed posts are still the source of truth; this page is the indexable summary of what June proved.
200K/8 on the dual DGX Spark pair, with Hermes capped at 200000. Fresh probes measured 209.66 tok/s at c8 short generation and 206.56 tok/s across a 24-request short soak. The attempted 200K/16 profile did not reach API startup, so 200K/8 is the sane current profile.200K/16 profile usable. See July DS4-F Testing for the fresh c16 benchmark and the still-failed 1.5M/C12 lane.| Track | Result | Why it mattered |
|---|---|---|
| Initial dual-Spark deployment | DeepSeek V4 Flash served across the two DGX Sparks with tensor parallelism, tool calling, thinking mode, and persistent launch scaffolding. | Proved the model could be made operational on the GB10 pair instead of staying a screenshot target. |
| Aiden production-v2 profile | The Aiden/B12X profile at 393216 context beat the 1M profile for Hermes-style concurrency in the June selection sweep. | Hermes needs enough context plus multiple moving agents; 1M as a default paid too much scheduling overhead. |
| Hermes operational probes | Plain chat, native OpenAI tool_calls[], long-context sentinel recall, c6 exact-response smoke, and streaming all worked. | This is the route shape Hermes actually cares about, not just an offline benchmark. |
| Tool-discipline testing | tool-eval-bench scored 87 / 100, with strong tool selection and structured output but weaker adversarial/state behavior. | Good enough for normal local agent work; not something to trust blindly with hostile retrieved context. |
| Terminal-Bench attempt | No public score published for the Hermes-tuned profile. | The endpoint was healthy, but Terminus-2's strict JSON completion protocol was the wrong harness fit for that route. |
These are fresh operational probes from the recovered DSpark/vLLM route on Spark1+Spark2. They are not leaderboard scores; they are the numbers that matter for Hermes routing: short-generation throughput, tool-call health, soak behavior, and the long-context boundary that stayed green.
| Profile / probe | Measured result | Decision |
|---|---|---|
1M/2 capped route | c1 63.10 tok/s, c2 91.11 tok/s, c4 94.42 tok/s; 250,417-token retrieval passed at d50. | Stable but under-concurrent for default Hermes traffic. |
384K/4 | c1 60.91 tok/s, c2 79.16 tok/s, c4 138.78 tok/s; c4 staggered 98.4 tok/s. | Throughput win, but ~250K d50 retrieval failed reproducibly; cap lowered. |
200K/16 | KV profiling reported 1,082,964 tokens and only 5.41× max concurrency for 200K requests; API never came up. | Do not use yet. It needs startup/debug work. |
200K/8 current route | c1 63.18 tok/s, c2 92.83 tok/s, c4 139.20 tok/s, c8 209.66 tok/s; c8 staggered 145.4 tok/s. | Current best production Hermes profile. |
200K/8 soak + retrieval | 24/24 short-soak requests OK at 206.56 tok/s; 200K target needle 3/3 PASS at ~139,188 actual tokens; 275K d50 boundary PASS at 191,328 actual tokens. | Keep Hermes spark-ds4 capped at 200000. |
Provenance: local artifacts under ~/.hermes/work/ds4-f-recovery/: bench_capped_route_20260701T144611Z.json, bench_capped_route_20260701T153645Z.json, bench_capped_route_20260701T165109Z.json, bench_200k8_c8_soak_20260701T165153Z.json, and staggered_200k8_c8_20260701T165227Z.log.
| Measurement | Published value | Source |
|---|---|---|
| Aiden Hermes context | 393216 | Aiden recipe |
| c6 concurrent short decode | 105.09 aggregate tok/s | Aiden benchmark probes |
| Tool benchmark | 87 / 100, 120 / 138 points | three-light benchmark read |
| llama-benchy concurrency | 69.1 total gen tok/s at c4 | llama-benchy section |
| Long context operational probe | 60,030-token sentinel recall passed | Hermes probes |
| July 1 current route speed | 200K/8: c8 209.66 tok/s; 24-request soak 206.56 tok/s; retrieval green to 191,328 actual prompt tokens | Local DSpark recovery artifacts, July 1 |
The June DS4-F work was a pivot away from maximum-context theater. The 1M profile launched, but the production question was different: which profile keeps real agents moving? The answer was the Aiden Hermes profile: large enough context, native tools, B12X/MTP serving, and better c6 behavior than the 1M route.
The cracks were equally useful. DS4-F is not immune to precision failures, crowded toolsets, adversarial search results, or long-lived state traps. The right deployment posture is: default local workhorse, guarded autonomy, and separate strict profiles for benchmark protocols.
This archive entry summarizes already-published June measurements and the July 1 DSpark speed probes. The July 1 table is operational route evidence, not a leaderboard benchmark.