Fetching from the wire…
01
02
03
04
05
06
07
08
Vibe Coding2026-09-22 · source-backed
Nimble takes text plus a schema of choice, boolean or score questions and returns typed answers with per-option probabilities, and cannot write free text at all. Its contribution over the other open typed-decision projects is the training recipe plus published numbers that don't flatter it: contrastive data curation builds paired examples differing in one critical fact so the correct answer flips, yielding 2,676 curated examples across 10 subject categories. On 324 held-out examples Bespoke-Nimble-9B agreed with reference labels 90.12% of the time against 66.36% for the Qwen3.5-9B base and 93.21% for Jev. Median 106ms per example on an H100. (GitHub)
Each link below shares sources, entities, or timing with this story.
Bespoke took a $40M Series A (Wing VC and 8VC, with angels including Jeff Dean) to build simulation-and-evaluation environments where agents learn and get tested before production. The bet: the missing layer isn't a bigger model, it's the harness that makes agents dependable....
iApp Technology released OpenThai-SystemOne September 20, a 0.8B Apache-2.0 model with weights and the full training recipe, explicitly to open the architecture TypeSafe kept closed. On Bespoke Labs' 13-subset benchmark with identical subsets, splits, instructions and sampler,...
@madiator's Bespoke Nimble uses contrastive data curation to lift the base model from 66% to 90% against Jev's 93% on the same task. Latent Space That's the most informative data point in the whole clone wave, because it suggests most of the accuracy is reachable with a public...
Compaction is where long sessions go to die. The model summarizes what happened, the summary drops the exact string you needed three hours later, and you don't find out until the agent confidently references a file path that never existed. It's the biggest source of silent con...
The September 5 release adds beam search via a beam_width request parameter returning the n best sequences, though it doesn't yet combine with speculative decoding, disaggregation, DP attention or HiCache. DeepEP v2's fixed-size ElasticBuffer engine as --moe-a2a-backend deepep...
The abliteration tool gained 215 stars to reach 30,103, but the stronger signal is downstream: the HF trending endpoint returns DavidAU/Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NM-DAU and Momoking/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4, both naming the too...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.