AI Daily Insights No. 92 · November 4, 2026
← Back to company guide · DeepSeek

DeepSeek models, Harness and open-source product timeline

Model versions, tool APIs and agent applications affect capabilities, billing and workflows in different ways. This guide connects our coverage by product and preserves each release, preview, prototype or planning stage.

Editorial review · Coverage through

News updates with each issue. Editorial reviews carry their own dates.

Progress in context

This editorial review dated November 4, 2026 covers a V4.1 Flash update, Search and Fetch APIs, the Harness plugin environment and the Super-Agent research prototype. Coverage of V5 concerns test-code references and reported pretraining plans. An official announcement is still needed to establish a release date.

Selected milestones

Dates refer to our coverage. Each milestone retains the reported stage at that time. Follow the news timeline for subsequent developments.

  1. V4.1 Flash update

    DeepSeek updates V4.1-Flash-1017, improving project development and shortening task completion time

    Read original coverage ↗

  2. Search and Fetch APIs

    DeepSeek opens Search and Fetch APIs at RMB 5 per 1,000 successful requests, with existing tool calls billed from November

    Read original coverage ↗

  3. V5 names in compatibility tests

    DeepSeek V5 names appear in test code without establishing a launch timetable

    Read original coverage ↗

  4. Super-Agent research prototype

    DeepSeek open-sources a Super-Agent prototype for team coding workflows

    Read original coverage ↗

  5. Harness Creator Mode

    DeepSeek showcases Harness Creator Mode for building reusable plugins through chat

    Read original coverage ↗

  6. V5 pretraining plans reported

    DeepSeek V5 plans Ascend pretraining as its Huawei collaboration extends deeper into infrastructure

    Read original coverage ↗

Common questions

How does Harness relate to model versions?

Harness provides the environment for files, execution, tools and plugins, while the model handles understanding and generation. Creator Mode builds and modifies reusable plugins. The product remains in preview iteration in this coverage.

Read original coverage · 2026-11-04 ↗

Is Super-Agent a finished product?

The official repository cited in our November 2 coverage labels it a research prototype. It connects a code repository with team collaboration, requires permissions and runtime configuration, and leaves parts of its design unimplemented.

Read original coverage · 2026-11-02 ↗

Does V5 have a confirmed release date?

In the reviewed coverage, the names reported on October 29 came from compatibility tests, while November 4 coverage attributed pretraining plans to supply-chain sources. DeepSeek had not publicly confirmed the training arrangements or release date.

Read original coverage · 2026-11-04 ↗

Related coverage timeline

DeepSeek showcases Harness Creator Mode for building reusable plugins through chat
Generated results can persist as reusable tools, allowing users to accumulate agent capabilities suited to their own work.
DeepSeek V5 plans Ascend pretraining as its Huawei collaboration extends deeper into infrastructure
Ascend adaptation now reaches compilation, computation and communication; training capability depends on sustained coordination across hardware and cluster software.
DeepSeek restores normal service after yesterday’s web and API disruption
As models enter continuously running applications, availability and response reliability shape the experience alongside reasoning capability.
DeepSeek Harness showcases Creator Mode for building plugins through conversation
Turning recurring needs into reusable plugins lets an assistant gain value as a user's working routines accumulate.
DeepSeek open-sources a Super-Agent prototype for team coding workflows
Once coding agents join a team, authorization, context continuity and notification frequency directly affect how much human attention they save.
DeepSeek V5 names appear in test code without establishing a launch timetable
Open engineering code can reveal product mechanisms, but a test identifier remains separate from model training, product definition and service availability.
OpenRouter handles 188 trillion tokens in a week as DeepSeek leads its model chart
API aggregators make switching easier, allowing capability, price and availability to appear more quickly in platform traffic.
DeepSeek and MiniMax complete registration and related open releases and overseas licensing
Registration progress should be read alongside product-specific information; actual deployment depends on model versions, licenses and service terms.
DeepSeek open-sources DeepInfer, the core engine behind large-scale V4.1-Flash inference
Open inference engines let third parties reuse system-level optimization, making deployment economics increasingly sensitive to caching, scheduling and real request patterns.
DeepSeek open-sources DeepInfer to share its inference framework and deployment techniques
Open weights make a model accessible; opening inference engineering further reduces the work required to serve it continuously to users.
DeepSeek 1017 leads open models on Arena WebDev with a latest score of 1664
Web generation is becoming more accessible, while model selection still needs working outputs and revision costs alongside leaderboard scores.
DeepSeek updates V4.1-Flash-1017, improving project development and shortening task completion time
Coding models increasingly compete on the cost of the whole task, with fewer repairs bringing them closer to the bill users actually pay.
DeepSeek opens Search and Fetch APIs at RMB 5 per 1,000 successful requests, with existing tool calls billed from November
An agent's total cost increasingly depends on the tools it invokes and how often it invokes them, beyond the model's token prices.
DeepSeek upgrades to V4.1-Flash-1017, switching its web, app and API to the new model
Model upgrades are increasingly sold on getting the same answer faster and cheaper rather than better answers, which matters more for agents than benchmark scores.
DeepSeek merges four teams into a Product Department led by Harness chief Cui Tianyi
As gaps between models narrow, turning a model into a product people enjoy using is becoming the new dividing line between labs.
DeepSeek goes down for nearly five hours and refunds seven days of API spending as credits
APIs are now production infrastructure for many companies, so model providers compete on uptime and accountability as well as price and capability.
Microsoft finishes pre-training MAI-Base-2, a 2.8-trillion-parameter model it says is 4x to 12x more compute-efficient than DeepSeek-V4-Pro-Base
Companies building their own large models now pitch "we didn't distill anyone" as a feature, adding a "where did the data come from" test to enterprise procurement that open models do not always pass.
DeepSeek refunds all API users seven days of spending after restoring service and says it has removed single points of dependency
When developers bet their products on someone else's model API, providers compete not only on price and capability but on how they make things right when something breaks.
DeepSeek Harness builds a Claude Code Mods compatibility layer in two days that runs three sample extensions
An extension interface for coding agents had a compatibility layer two days after it appeared, which suggests this layer makes a poor moat and the first mover only wins the chance to define the standard.
DeepSeek and Huawei open-source Ascend programming tools, filling a key gap for training on Chinese chips
The hard part of chip independence was never just making the chip but getting developers to write code for it, and DeepSeek is solving that toughest software problem for Huawei.
DeepSeek Harness v0.2 preview ships desktop installers, taking the agent tool off the command line
Competition among agent tools is shifting from whose model codes better to whose installer is easier and whose plugins are easier to find; the desktop is the new battleground.
Liang Wenfeng tells investors DeepSeek is training a 2-trillion-parameter model and eyeing 8 trillion
The lab famous for saving compute is now quoting trillions of parameters; Chinese model competition has switched its unit from efficiency to scale.
DeepSeek lets V4.1 Flash carry a generational change, as a leak points to a 3T-parameter Code 2.0 in September
Version numbers no longer mark a capability jump; unit cost is the column that has to change.
DeepSeek V4.1 Flash Exposes Day-One Support Gap as Nvidia Runs vLLM Across Six GPU Families
New-model support has become an hourly contest; a usable day-one image can influence developer choice as much as hardware specifications.
DeepSeek App Adds Four Reading Voices, While Talk of a New TTS Model Remains Speculation
Four new voices are a confirmed product change; there is no evidence yet that the underlying model is new, so the feature and a model release must remain separate.
DeepSeek Launches V4.1 Flash, Using a 552-Billion-Parameter Architecture to Cut Inference Costs
DeepSeek is debuting its new architecture in a low-priced Flash model, leaving real-world task costs to determine whether the efficiency gain holds.
DeepSeek Harness 0.1.5 Arrives With Model Optimization and Expanded Plugin File Workflows
DeepSeek is pushing model upgrades into the agent runtime at the same time, giving developers a broader product interface alongside the new model.
DeepSeek plans to route V4 Pro requests to V4.1 Flash after launch and charge Flash rates
Backend substitution reduces migration work but leaves developers needing version identification and regression tests. Lower bills and stable task performance need to be accepted together.
DeepSeek opens a limited V4.1 Flash beta, claiming a new architecture and native multimodality without a full technical release
The limited beta offers a window to test the new architecture; improvements in capability, speed and cost each need measurement on the same real tasks.
DeepSeek open-sources V4-Flash-Vision-Exp weights, 305B parameters under an MIT license
305B weights under MIT put the first squeeze not on OpenAI but on every domestic multimodal API billing by the call.
DeepSeek First Seven Months: $70.7M Revenue, API Gross Margin Hits 82.9%
The 82.9% API gross margin proves the inference business itself can make money; the question has always been who pays for the free half.
NVIDIA Buys Poolside's Technology and Team for $7 Billion, Builds Its Own Open-Source Model to Rival DeepSeek
A chipmaker building the strongest open-source model itself was never about selling the model — it's about becoming the default option that routes everyone's inference onto its own hardware.
DeepSeek Adjusts API Billing — Entire Weekend Now Billed at Off-Peak Rate
Where the pricing sheet loosens is usually where overcapacity first surfaces.
DeepSeek Launches Experimental Multimodal Model V4-Flash-Vision-Exp, Reading Up to 600 Images per Request
Comparison numbers produced on a self-run Harness are only worth as much as how quickly third parties can reproduce them.
DeepSeek Harness Ships rc.8, Adding Image Input and Installable Sub-Agents
Converting images to text and feeding the result to text-only models paves the way with engineering before model capabilities catch up.
DeepSeek API Adopts Peak/Off-Peak Pricing; V4 Pro Peak Output Rises to 27 Yuan per Million Tokens
When APIs start billing by peak and off-peak windows, compute officially becomes a utility.
DeepSeek Open-Sources Agent Framework Harness, Hits 95K GitHub Stars in Two Days
The real objective of open-sourcing a runtime is getting others' engineering habits to take root in your abstraction.
DeepSeek open-sources agent framework Harness: every component is pluggable
Beyond the model, the agent runtime is becoming the next layer to be conquered by open source.
DeepSeek Releases V4-Pro at One-Seventh the Price of Kimi K3
List price cut to a tenth, cache prices up tenfold — DeepSeek has turned a price war into a price-structure war and shifted the full burden of price comparison onto developers.
DeepSeek open-sources agent framework Harness with an "everything is a plugin" philosophy
With models free and frameworks open-sourced, the truly scarce assets left in the agent race are usage entry points and dominance of the plugin ecosystem.
DeepSeek Quietly Lists V4-Pro-0813 with 1M Context and 384K Max Output
The four-times-cheaper flagship model carries its own price-increase warning; cost models need to leave a line for it.
DeepSeek Registers 'DeepSeek Harness Team' Official Account Under Beijing DeepSeek
Between shipping a model and shipping a product sits an organization that can interview every day and open an official account.
DeepSeek V4-Flash 0731 Third-Party Benchmarks Are In: Just 13B Active Parameters, Agent Score Beats Its Own Pro Edition
Redoing post-training once was enough to overtake the premium version, showing the bar for agent capability is shifting from "stacking parameters" to "refining the recipe" — precisely the stage where the open-source camp is closing the gap fastest.
Unitree Technology's strategic placement list includes DeepSeek; Wang Xingxing says it will receive model architecture and intelligent-computing cluster support
The easiest way for a hardware company to get a brain is to make the brain-maker a shareholder.
DeepSeek Backend Notice Flags Major API Price Hike, Hitting an Inflection Point in the Low-Price Era of Chinese Large Models
The price slasher is raising prices itself — the ledger behind rock-bottom pricing no longer balances.
DeepSeek's Updated V4-Flash Matches GLM-5.2 in Benchmarks at One-Tenth the Price
When benchmark parity costs a tenth of the price, the middle-tier API has no story left to tell.
Artificial Analysis Estimates: DeepSeek V4-Flash Costs 3 Cents per Task
Per-million-token pricing is losing its reference value; what counts is the total bill for getting a task done.
DeepSeek V4-Flash official release goes live, output price cut to 2 yuan per million tokens — one-third of Pro
The moment a cheap model catches up to a premium one, what gets eliminated isn't the premium model — it's model routing as a business.
DeepSeek Ships V4-Flash Stable: Weights Open-Sourced Same Day, Agent Benchmarks Overtake Its Own Pro Preview
Run the same model twice, and the second pass is worth 10 extra points — post-training is becoming a more valuable asset than parameter count.