Coding Assistants: A Volatile Market with Uncertain Pricing

Remarkably volatile. That is how Gartner portrays the market for coding assistants.

It is the second Magic Quadrant Gartner has dedicated to this space, continuing from the broader one focused on “cloud AI services for developers.”

Compared with last year, the mandatory functional elements have evolved little. In broad terms, the offering needed to provide:

• Multiline auto-completion with natural-language suggestions
• Generation of unit tests and documentation
• General-purpose assistance across multiple languages and frameworks
• Awareness of the context of the company’s code repositories and other sources such as documentation
• Integrated chat within the development environment, with the ability to reference files and folders
• Assurance that client code or documentation will not be used to train models
• Administrative options such as excluding code and setting programming styles

Gartner nonetheless took into account the shift in the market, marked by a move away from IDE plug-ins toward autonomous environments. It also incorporated the agent-based aspect into its assessment.

As for market volatility, the firm points to Windsurf (formerly Codeium) as evidence. OpenAI attempted to acquire it, but Cognition ultimately took it over, with the leadership team joining Google.

Another illustration: the controversy Cursor generated by raising its prices in response to higher costs of its “raw material” — Anthropic’s models. The latter poses an even bigger threat as it has positioned itself in this market with Claude Code. It signals a broader trend among LLM providers to build their own code assistants.

14 providers, 5 “Leaders”

Reflecting this volatility, the ranking shifts quite notably along each axis of the Magic Quadrant. Of the 14 vendors listed, four are new entrants. One of them sits among the “Leaders”: Anysphere, the publisher of Cursor.

On the “Execution” axis, which is meant to translate the ability to actually meet demand (customer experience, pre-sales performance, product/service quality, etc.), the situation is as follows:

On the “Vision” axis, which captures strategies (sector, geographic reach, go-to-market, product, etc.):

Deployment and customization options limited at Amazon

Amazon is less exhaustive than other providers in terms of how many IDEs it supports, Gartner noted last year. A shortcoming that has been resolved: today, Amazon Q Developer is praised for its diffusion across numerous environments. It also scores well on the agent-based front, particularly in modernization. Added to that are investments in automated reasoning, already highlighted last year. The same goes for collecting customer feedback. Amazon also benefits from the quality of its operational support (security controls, zero-downtime updates, availability of account managers).

Compared with the other hyperscalers ranked here, Amazon undertakes fewer commercial efforts focused on coding assistants. Its business model can also complicate cost planning and benchmarking. And customization options (fine-tuning, prompt libraries, etc.) remain limited, as do deployment options. More broadly, the offering is optimized for the AWS ecosystem, which can incur costs if you switch providers. The Kiro IDE is a workaround, but it was still in an experimental phase at the time Gartner conducted its assessment.

Cognition, less mature on the commercial approach

Cognition earns points for its thorough support of programming languages and the robustness of its IDE integrations (in addition to having its own). Gartner appreciates the product’s capabilities in code customization, testing, debugging, and documentation generation. It also notes the flexibility provided by integrating agent-based capabilities within the development cycle. Positive notes also for the Cortex reasoning engine and the ease of deployment (hybrid models).

Hard to ignore the doubts about Windsurf’s viability after the leadership left for Google, and, in parallel, about its business model as agent integration goes hand in hand with a transition from per-seat licensing to usage-based pricing. Gartner also flags a lack of resources that limits Cognition’s ability to expand beyond dev-centric markets. Overall, the commercial approach remains less mature among the other Leaders, both in terms of channel and verticalization.

Complex pricing at GitHub

GitHub Copilot excels at core tasks (code translation, runtime understanding, automatic documentation), according to Gartner. Another strength: the level of integration into the SDLC (native availability in GitHub and Visual Studio Code, among others). Strengths also include governance and integration with codebases.

Gartner notes that sector-specific adaptation of the offering is minimal. It also points to pricing complexity. The multiplicity of SKUs as well as the interaction with GitHub Enterprise subscriptions contribute to opacity. It is not always easy to align GitHub’s contractual terms with Microsoft’s in cross-platform offerings.

GitLab, not ahead of the market

Last year Gartner praised GitLab for its marketing strategy, as well as the visibility and engagement that followed. It also appreciated the amount of resources dedicated to development and support. That remains the case, which explains the continued reliance on indirect customer acquisition. Other strengths include the level of integration within the GitLab platform and deployment flexibility.

GitLab is less well positioned than rivals when it comes to factoring in enterprise context. It is generally not ahead of the market on several components (chat, analytics dashboards, self-hosting options). Its messaging tends to center on “AI for DevOps” rather than the core functionality, risking diminished recognition among developers.

Translation, documentation… Google lagging on several uses

Gartner appreciates the seamless integration of Gemini Code Assistant into IDEs and the Google Cloud Platform ecosystem. A plus is also the pricing approach, with a free individual tier that helps attract a user base. Other praised aspects include the quality of support, the robustness of SLAs, and the bridges with DeepMind, which support adapting models to real-world use cases.

Google lags behind the other Leaders in code translation, runtime understanding, and documentation generation. It also trails in model selection and in tailoring its offering to industries and company sizes. Pricing lacks transparency, particularly regarding discounts.