Recall
Handwriting and screenshots become searchable cards, and the extension reaches Safari
A side-by-side editorial comparison of Ollama and ragnar — release velocity, themes, recent moves, and the top alternatives to consider.
A release train of small runtime wins between model drops
Ollama is in the gap between model launches, spending its releases on per-request overhead and desktop polish rather than new capability. The v0.32.15 train adds a model metadata cache to cut per-request cost, an onboarding flow for the desktop app, and a temporary MLX-C patch carried in-tree. The substantive model work in this window is still Qwen 3.8 27B at v0.32.12, with its Apple Silicon MLX build.
ragnar turned its RAG store into an MCP server, so coding agents can search it directly.
ragnar builds retrieval-augmented generation stores in R on DuckDB, handling document chunking, embedding, and hybrid vector plus BM25 retrieval, and registering itself as a tool for ellmer chats. Version 0.3.0 adds mcp_serve_store(), which exposes a store over MCP to local clients such as Codex CLI and Claude Code, alongside Azure AI Foundry and Snowflake Cortex embedding providers. Store version 2, introduced in 0.2.0, brought chunk deoverlapping on retrieval and automatic heading augmentation.
Ollama is in the gap between model launches, spending its releases on per-request overhead and desktop polish rather than new capability. The v0.32.15 train adds a model metadata cache to cut per-request cost, an onboarding flow for the desktop app, and a temporary MLX-C patch carried in-tree. The substantive model work in this window is still Qwen 3.8 27B at v0.32.12, with its Apple Silicon MLX build.
The shape is consistent: a headline model addition every few weeks, then a run of releases tightening the runtime around it — quantization paths, prefill speed, renderer fixes. Desktop is quietly becoming a first-class surface rather than a wrapper on the CLI, and the MLX path keeps getting hand-tuned for Apple Silicon ahead of the generic backend.
Expect the next headline release to be another model addition with a paired MLX build, since that is what four of the last several notable entries look like, with the release-candidate tags continuing to carry the user-visible desktop work ahead of the final tag.
ragnar builds retrieval-augmented generation stores in R on DuckDB, handling document chunking, embedding, and hybrid vector plus BM25 retrieval, and registering itself as a tool for ellmer chats. Version 0.3.0 adds mcp_serve_store(), which exposes a store over MCP to local clients such as Codex CLI and Claude Code, alongside Azure AI Foundry and Snowflake Cortex embedding providers. Store version 2, introduced in 0.2.0, brought chunk deoverlapping on retrieval and automatic heading augmentation.
The package keeps widening who can reach a store and how many ways they can query it. Retrieval accepts vectors of queries, the ellmer tool withholds chunks it has already returned so an agent can dig deeper across calls, and now the store is reachable from outside R entirely. Embedding providers are added steadily — LM Studio, then Azure and Snowflake — which keeps the store portable across whoever supplies the vectors. Breaking changes are accepted readily at this stage, including a renamed default tool prefix and a flipped default in ragnar_find_links().
More MCP surface is the natural next step now that serving exists, since the retrieval tool already has the multi-query and no-repeat behavior that agent-driven search depends on.
Other ai-assistants products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Ollama or ragnar.
Handwriting and screenshots become searchable cards, and the extension reaches Safari
Evaluation content dominates a feed whose real move was handing agents the admin panel
ClearML is filling in the hyperdataset lifecycle while hardening the SDK against what it loads.
Baseten is selling to the labs that build models, not just the developers who call them.
Between a BTS tie-in and free student plans, Gemini quietly moves into a Waymo
Perplexity is selling access to other people's models, and now repricing them weekly.
See all Ollama alternatives → · See all ragnar alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Ollama is currently shipping more aggressively (velocity 5.0 vs 0.0), with 0 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Ollama is currently shipping more aggressively (velocity 5.0 vs 0.0), with 0 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other ai-assistants products to evaluate alongside.
Top Ollama alternatives in ai-assistants are ranked by recent ship velocity. Browse the "Ollama alternatives" section above for the current picks, or visit /alternatives/ollama for the full list with editorial commentary on each.
Top ragnar alternatives in ai-assistants are ranked by recent ship velocity. Browse the "ragnar alternatives" section above for the current picks, or visit /alternatives/ragnar-r for the full list with editorial commentary on each.