Plan event · 29 Sept 2026
All plan events · 31 providers tracked
Neural Wattmodels
What moved
8 canonical models on Basic / Standard / Pro: ['DeepSeek V4 Flash','GLM-5.2','Kimi K2.7 Code','Gemma 4 31B','Qwen3.6-35B','GLM-5.3','Kimi K3','Qwen 3.8 27B'], '+6 more' variant rows → 10 canonical models on Basic / Standard / Pro / Max: GLM-5.2 replaced by GLM-5.3 (same published rate $1.45/$4.50) and DeepSeek V4.1 Flash, GLM-5.3-Flash and MiMo V2.6 Pro added; '+6 more' → '+4 more' variant rows
Neural Watt roster sighted on the 2026-09-29 pricing dump (GLM-5.2 is gone from both the estimator table and the public /v1/models catalogue), logged as one mixed event: a cut (GLM-5.2 out) plus three additions, with the roster as a whole moving 8 → 10 canonical models on every plan. GLM-5.2 is recorded as removed from the lineup, not a discontinuation — the model stays in the registry and other providers still serve it.
Verified 2026-09-29 · source
Affected plans
More Neural Watt events
- 2026-09-11 · Models — Neural Watt pricing page grew to 14 models; GLM 5.3 ($1.45/$4.50, 1M context, exited previ
- 2026-09-23 · Price — Calculator cache correction: Neural Watt publishes no cached-read rate, so cached tokens n
- 2026-09-29 · Launch — Neural Watt launched a fourth self-serve tier, Max $200/mo with 28.57 kWh of monthly energ
- 2026-09-29 · Price — Recalibration of Neural Watt's energy-derived calculator rates: wh_per_mtok is a trailing-