SCANNING● LIVE DATAUTC 2026-06-29 00:00:00
VENDORS 23RECEIPTS 32CRITICAL 57D 0 NEWv0.9.4-rc1
gotnerfedgotnerfed
✓Filed by gotnerfed.comCase #D-2023-863Classification CLEAN-trackOpened 2026-02-08Last scanned 2026-06-29 01:00 UTCACTIVE
AI CHAT/DeepSeek/founded 2023

DeepSeek

Open-weight reasoning model API

2 receipts logged, none critical.

Defining receipt|DeepSeek ships V4 and sets a hard Jul-24 retirement for the deepseek-chat / deepseek-reasoner model names|2026-04-24
► Clean5-24 · no critical receipts; nothing on watch.
Ranked #10 of 23 tracked vendors.
Receipt trail
2 filed · latest 66d ago
◆ At a glanceAll time · since 2026-02-08
66days
Since last receipt
DeepSeek ships V4 and sets a hard J… · minor
2
Receipts
all time
0
Critical
—
0
Major
none on record
1
Last 90d
watching
20
Peak nerf
Apr 24 · model swap
~75d
Cadence
avg between receipts
►Verdict: Clean
Receipt trail
2 filed
Methodology ↗
↗ Current pricing pageHow the index is calculatedEmbed badgeProcurement report (PDF)Compare to peers
◆ 30 / 60 / 90 forecast →
likely next: model swap · expected severity minor
Next 30d49%
prob of any change
Next 60d74%
prob of any change
Next 90d87%
prob of any change
Observed 2 receipts over 75 days (cadence ~38d). Last receipt 66 days ago. Most common kind: model-swap. Brier-scored against subsequent reality on the public leaderboard.
Vendor average35%Category avg27%Brier · last 90d0.18Modelv0 heuristic
◆ Nerf Index · trailing 12 months▲ +12 vs. 90 days ago · now 19
JunAugOctDecFebAprJun
Recency-weighted Nerf Index, 30-day rolling. Dots mark receipts. Index breached “clean” threshold on 2026-04-24 at the defining receipt.
◆ Community sentimentReddit · last 90d

Community sentiment

No sentiment data yet. Reddit posts about DeepSeek are scored by Claude Haiku 4.5 on first render and cached daily - the next page load will populate this card. Requires ANTHROPIC_API_KEY + Vercel KV.

Net = (positive − negative) / total Reddit posts mentioning DeepSeek in the trailing 7 days. Updated daily. Read methodology →
Receipts · all time · reverse chronological

Every silent change DeepSeek made.

2 receipts · 0 critical · 0 major · 2 minor
2 MORE RECEIPTS HIDDEN · LAST 12 MONTHS

You’re seeing the last 30 days. Personal sees the last year.

Free tier shows the most recent 30 days of receipts per vendor. Personal ($9/mo) unlocks the full 1-year diff history - plus a 3-vendor watchlist, real-time alerts on every severity, migration playbooks, and Slack/Discord/Teams webhooks.

◆ Escape pod · if you're leaving

Free & open-source alternatives to DeepSeek.

See full migration playbook ->
◆ ESCAPE POD · FREE / OPEN-SOURCE ALTERNATIVES
5 ranked alternatives to DeepSeek.
  • Ollama ↗FREE / OSSMIT★ 130k
    Easiest way to run a local LLM. Pair with Open WebUI for full ChatGPT replacement.
    Run Llama/Mistral/Gemma/DeepSeek locallyOllama-compatible API for any clientBrew/winget/curl install
    install: brew install ollama && ollama run llama3.3
    caveat: Needs 16GB+ RAM for usable models. GPU strongly recommended.
  • Open WebUI ↗FREE / OSSMIT★ 95k
    The default ChatGPT-replacement self-host. Massive ecosystem.
    ChatGPT-style web UI for any backend (Ollama, OpenAI-compatible)RAG with file uploadsFunction calling, web search plugin
    install: docker run -p 3000:8080 ghcr.io/open-webui/open-webui:main
    caveat: Best paired with Ollama running locally.
  • Permissive (MIT) open-weight model targeting ChatGPT/Claude-class chat and agentic coding without API lock-in - the strongest open option for a self-hosted stack.
    Fully-open (MIT) flagship MoE from Z.ai / Zhipu AI (weights released Jun 2026)~744B total / ~40B active parameters, 1M-token context (~5x GLM-5.1)Strongest open-weight coding model at release - #2 on Code Arena, ~#3 on FrontierSWE behind only Fable 5 / Opus 4.8Tool use + reasoning, served via vLLM/SGLang
    install: ollama run glm-5.2 (or download weights from HuggingFace: zai-org/GLM-5.2, serve via vLLM/SGLang)
    caveat: MIT, fully open, no usage restrictions. At ~744B params it is multi-GPU / data-center scale - use a hosted provider if you lack the hardware. Note: the hosted Z.ai API routes through China (a data-residency consideration); self-hosting the open weights avoids that.
  • Drop-in open-weight replacement for the OpenAI/Anthropic chat APIs. Flash is the locally-runnable variant; serve it behind Open WebUI.
    MIT-licensed open-weight MoE: V4-Flash (284B/13B active) + V4-Pro (1.6T/49B active)1M-token contextNear-frontier quality at a fraction of API costRun locally or self-host the backend
    install: Download from HuggingFace (deepseek-ai/DeepSeek-V4-Flash); serve via vLLM/SGLang
    caveat: Even Flash (284B) needs serious VRAM/multi-GPU; Pro (1.6T) is data-center scale. MIT license, weights fully open.
  • Unrestricted (Apache-2.0) open-weight model to replace the OpenAI/Anthropic/Gemini chat APIs in a self-hosted stack.
    Apache-2.0 open-weight MoE from Mistral AI: 675B total / 41B active256K context, multimodal (text + vision)Production-grade general-purpose modelServe via vLLM/SGLang or run through Ollama
    install: ollama run mistral-large-3 (or download weights from HuggingFace: mistralai/Mistral-Large-3-675B-Instruct-2512-BF16, serve via vLLM/SGLang)
    caveat: Apache-2.0, no usage restrictions. December 2025 release. At 675B params it is multi-GPU / data-center scale - use a hosted provider if you lack the hardware.
List manually curated · last reviewed 2026-05-13Full picker →