News·SN96 Verathos GitHub·Sep 4, 2026, 15:40 UTC
Verathos v0.2.0 – Sleipnir: Verifiable Multi-Node Model Serving
About SN96 Verathos
Verathos v0.2.0 – Sleipnir: Verifiable Multi-Node Model Serving Verathos v0.2.0 expands SN96 from independently operated vLLM endpoints to a shared serving layer that also supports Sleipnir coordinated GGUF mesh pools. Builders continue to use one OpenAI-compatible API while independent operators can combine qualified CUDA capacity to serve larger and more varied open, subnet-trained, and fine-tuned models. What changed Adds Sleipnir coordinated mesh pools with guided coordinator and worker setup, authenticated membership, lifecycle controls, status probes, and operator dashboards. Adds verifiable mesh inference with pinned execution snapshots, signed receipts, independent validator checks, and unpredictable capacity audits. Prioritizes deadline-bearing capacity receipts ahead of background delivery work on multi-GPU mesh workers for more reliable audit completion. Keeps relaunched proof workers out of service until their authenticated proof-weight cache has converged, preventing cold or partial state from being admitted as healthy capacity. Unifies discovery, health, routing, sessions, billing, x402 payments, proof display, scoring, and validator snapshots across vLLM and mesh endpoints. Keeps healthy mesh routes discoverable during bounded generation-slot saturation while preserving request failover and retry signaling. Adds conviction-lock-based daily inference credits alongside account credits and x402 pay-per-request access. Introduces independent runtime-family emission allocation with explicit unallocated burn when a reserved family has no eligible serving capacity. Preserves the existing vLLM score and weight vector exactly while the mesh allocation is disabled. Enforces authenticated CUDA-only mesh admission on mainnet while retaining unsupported entries in diagnostics with clear remediation. Carries forward qualified software support for Qwen3.6 35B sparse-MoE and the existing FP8, INT4, GPTQ, and unquantized serving lanes. Refreshes the public documentati
Read the github release at SN96 Verathos GitHub ↗
SN96 Verathos at a glance
SN96, Verathos (α token ᚛), is one of the independent markets on the Bittensor network, each with its own α token and its own price. Its α currently trades at 0.006163 τ (≈ $1.40), against 3,074 τ of pool liquidity. On-chain, the best-delegated validator on it yields about 232.9% APR, emissions add 0.000391 τ to the pool each block, epochs run every 360 blocks. It was registered at block #7,692,872.
More from SN96 Verathos
Taosis indexes announcements, articles and releases about Bittensor subnets from their own channels and keeps a record of each. The text above is the source’s; the figures are read from the chain by Taosis.