Ia - #1

Open

Abraao320 wants to merge 3205 commits into

TheLocalLab:mainfrom

This pull request is big! We're only showing the most recent 250 commits

Commits on Mar 18, 2026

Fix `--extra-flags` breaking short long-form-only flags like `--rpc`
oobabooga
committed
2a6b1fd
llama.cpp: Support literal flags in `--extra-flags` (e.g. `--rpc`, `--jinja`)
oobabooga
committed
7e54e7b
Merge remote-tracking branch 'refs/remotes/origin/dev' into dev
oobabooga
committed
08ff3f0
Security: server-side file save roots, image URL SSRF protection, extension allowlist
oobabooga
committed
c8bb212
Security: server-side file save roots, image URL SSRF protection, extension allowlist
oobabooga
committed
256431f
UI: Fix the instruction template delete dialog not appearing
oobabooga
committed
fef2bd8
API: Remove leading spaces from post-reasoning `content`
oobabooga
committed
ca36bd6
Fix AMD installer failing to resolve ROCm triton dependency
oobabooga
committed
eeb0e57

Commits on Mar 19, 2026

Use `logger.exception()` instead of `traceback.print_exc()` for error messages
oobabooga
committed
779e761
Cleanup `modules/chat.py`
oobabooga
committed
dde1764
Remove ancient/obsolete instruction templates
oobabooga
committed
5453b9f
Minor cleanup across multiple modules
oobabooga
committed
e0e20ab
Merge branch 'main' into dev
oobabooga
committed
b9922f7
Reduce the size of portable builds by using stripped Python
oobabooga
committed
b3eb0e3

Commits on Mar 20, 2026

Update exllamav3 to 0.0.26
oobabooga
committed
843de8b
Minor cleanup
oobabooga
committed
2e4232e
API: Move OpenAI-compatible API from extensions/openai to modules/api
oobabooga
committed
bf6fbc0
API: Fix debug_msg truthy check for OPENEDAI_DEBUG=0
oobabooga
committed
1a91057
API: Handle --extensions openai as alias for --api
oobabooga
committed
8551419
API: Fix _start_cloudflared raising after first attempt instead of exhausting retries
oobabooga
committed
7c79143
Add missing __init__.py to modules/grammar
oobabooga
committed
f0e3997

Commits on Mar 21, 2026

API: Add Anthropic-compatible /v1/messages endpoint
oobabooga
committed
0216893
API: Use top_p=0.95 by default
oobabooga
committed
f2c9097

Commits on Mar 22, 2026

Update API docs to mention Anthropic support
oobabooga
committed
2c4f364
llama.cpp: Don't suppress llama-server logs
oobabooga
committed
9488df3
Follow-up to previous commit
oobabooga
committed
1dda5e4

Commits on Mar 23, 2026

Commits on Mar 24, 2026

Revert "Follow-up to previous commit"
oobabooga
committed
286bbb6
Revert "llama.cpp: Don't suppress llama-server logs"
oobabooga
committed
a7ef430
Update README
oobabooga
committed
c9d2240
Update llama.cpp
oobabooga
committed
5b8da15
UI: Minor polish
oobabooga
committed
5814e74
Fix GPT-OSS tool-calling after 9ec20d97
oobabooga
committed
7505026
UI: Minor polish
oobabooga
committed
f48a2b7
Merge pull request #7441 from oobabooga/dev
oobabooga
authored
dd9d254

Commits on Mar 25, 2026

Commits on Mar 26, 2026

Add ik_llama.cpp support via `--ik` flag
oobabooga
committed
4cbea02

Commits on Mar 28, 2026

Commits on Mar 29, 2026

ik_llama.cpp: Auto-enable Hadamard KV cache rotation with quantized cache
oobabooga
committed
0466b6e

Commits on Mar 31, 2026

Several small code simplifications
oobabooga
committed
6382fbe
API: Implement echo + logprobs for /v1/completions endpoint
oobabooga
committed
71c1a52

Commits on Apr 1, 2026

Update llama.cpp
oobabooga
committed
328534b

Commits on Apr 2, 2026

Fix ExLlamav3 OOM on prompt logprobs and qwen3_5_moe HF compat
oobabooga
committed
4073164
Don't pass torch_dtype to transformers, autodetect from model config
oobabooga
committed
a32ce25
API: Add token ids to logprobs output
oobabooga
committed
c10c6e8
API: Optimize prompt logprobs and refactor ExLlamav3 forward pass
oobabooga
committed
ea1f8c7
Add dedicated ik portable requirements files and remove macOS ik builds
oobabooga
committed
c50e17b
Update exllamav3
oobabooga
committed
8f8b57a
Update transformers
oobabooga
committed
6a1f720
Update accelerate
oobabooga
committed
468cb5c
Remove ik macOS wheels from full requirements
oobabooga
committed
80e81a5
Security: Fix SSRF in superbooga extensions
oobabooga
committed
f6f8f14
Fix top_logprobs_ids missing for llama.cpp loader
oobabooga
committed
091037e

Commits on Apr 3, 2026

Update llama.cpp
oobabooga
committed
a61bde5
Update the custom gradio wheels
oobabooga
committed
d841574
API: Improve cache clearing in logprobs
oobabooga
committed
7aab2fd
Merge pull request #7452 from oobabooga/dev
oobabooga
authored
ae699ac
Fix portable builds not starting due to missing ik element
oobabooga
committed
b108c55
Merge pull request #7453 from oobabooga/dev
oobabooga
authored
9374a4e
Add Gemma 4 tool-calling support
oobabooga
committed
6e2b70b
API: Add warning about vanilla llama-server not supporting prompt logprobs + instructions
oobabooga
committed
42dfcdf
Merge pull request #7455 from oobabooga/dev
oobabooga
authored
b11379f
llama.cpp: Disable jinja by default (we use Python jinja, not cpp jinja)
oobabooga
committed
a1cb5b5
Revert "llama.cpp: Disable jinja by default (we use Python jinja, not cpp jinja)"
oobabooga
committed
000d776
Fix crash when no model is selected (None passed to resolve_model_path)
oobabooga
committed
66d1a22
Update the custom gradio wheels
oobabooga
committed
8bba9ec
Revert "API: Add warning about vanilla llama-server not supporting prompt logprobs + instructions"
oobabooga
committed
95d6c53
Update llama.cpp
oobabooga
committed
131a9a0
Update the custom gradio wheels
oobabooga
committed
8e8e1ba
Update the custom gradio wheels
oobabooga
committed
6b66da8
Merge pull request #7461 from oobabooga/dev
oobabooga
authored
0050a33
Update the custom gradio wheels
oobabooga
committed
5fb8c4f
fix(security): sanitize filenames in all prompt file operations (CWE-22) (#7462)

oobabooga
and
Alex Chen
authored
8ecdb41
API: Fix tool call parser crash on non-dict JSON output
oobabooga
committed
fc35aca
Merge pull request #7463 from oobabooga/dev
oobabooga
authored
62e67ad

Commits on Apr 4, 2026

Add Windows + ROCm portable builds
oobabooga
committed
2fbaee5
Merge pull request #7464 from oobabooga/dev
oobabooga
authored
b5afecc
Cleanup modules/chat.py
oobabooga
committed
54b2f39
Update README
oobabooga
committed
16af11f
UI: Fix tool buffer check truncating visible text at end of generation
oobabooga
committed
e0ad4e6
API: Fix loader args leaking between sequential model loads
oobabooga
committed
9183dc4
API: Remove deprecated "settings" parameter from model load endpoint
oobabooga
committed
2eef90a

Commits on Apr 5, 2026

UI: Improve the hover menu looks
oobabooga
committed
7fed60f
UI: Improve message text contrast
oobabooga
committed
ffea8f2
UI: Improve scrollbars style
oobabooga
committed
41bce3f
UI: Improve message action icon visibility in light mode
oobabooga
committed
8cb7fe9
UI: Fix inline LaTeX rendering by protecting $...$ from markdown (closes #7423)
oobabooga
committed
1b403a4
UI: Reduce spacing above chat input
oobabooga
committed
0c033ca
UI: Make accordion outline styling global
oobabooga
committed
dfd8ec9
UI: Make table and hr borders more subtle
oobabooga
committed
ee917cd
UI: Soften message action icons in light mode
oobabooga
committed
e8b31c0
UI: Improve blockquote border width and color
oobabooga
committed
1f49a64
UI: Minor change
oobabooga
committed
91f9b01
Update the custom gradio wheels
oobabooga
committed
9805ddc
Image generation: Embed generation metadata in API image responses
oobabooga
committed
c63a79e
Simplify modules/image_models.py
oobabooga
committed
544fcb0
Pre-compile LaTeX regex in html_generator.py
oobabooga
committed
422f42c
Fix "address already in use" on server restart (Linux/macOS)
oobabooga
committed
d78fc46

Commits on Apr 6, 2026

Call ik portable build folders text-generation-webui-ik-version
oobabooga
committed
f8db23b
UI: Hide spin buttons on number inputs
oobabooga
committed
223dd4b
UI: Move cpu-moe checkbox to extra flags (no longer useful now that --fit exists)
oobabooga
committed
abc3487
UI: Add MCP server support
oobabooga
committed
b1d06dc
Fix image generation: default to SDPA attention backend
oobabooga
committed
05e4842
Minor change after b1d06dcf96e2b5958ae004b8c9bbb0fc8518328b
oobabooga
committed
7b2f15e
Follow-up to d78fc46114a4ce1de505fc286798372ddaa0c32d
oobabooga
committed
4d6230a
API: add instruction_template support to the model load endpoint
oobabooga
committed
c26ffdd
API: Fix IPv6 address formatting
oobabooga
committed
193424c
Fix GPT-OSS tag leak during streaming between thinking and tool calls
oobabooga
committed
cb51192
Fix crash when truncating prompts with tool call messages
oobabooga
committed
775c913

Commits on Apr 7, 2026

Update llama.cpp/ik_llama.cpp
oobabooga
committed
778e1c4
Remove hardcoded trust_remote_code=True in embedding loader
oobabooga
committed
e18f32c
Merge pull request #7471 from oobabooga/dev
oobabooga
authored
9dcf574

Commits on Apr 8, 2026

Update README
oobabooga
committed
5ad199e
Merge pull request #7478 from oobabooga/dev
oobabooga
authored
0c3720a

Commits on Apr 10, 2026

Update llama.cpp
oobabooga
committed
24cc0e2
Update README
oobabooga
committed
9bc3f9f

Commits on Apr 11, 2026

Reduce VRAM peak in prompt logprobs forward pass
oobabooga
committed
326867e
Update llama.cpp
oobabooga
committed
456afad

Commits on Apr 12, 2026

Update exllamav3
oobabooga
committed
e9d7feb
Fix bos/eos tokens not being set for models without a chat template
oobabooga
committed
add9161

Commits on Apr 13, 2026

Commits on Apr 14, 2026

Update README
oobabooga
committed
5f16b86

Commits on Apr 15, 2026

Update llama.cpp
oobabooga
committed
3dcb30b
Revert "Update exllamav3"
oobabooga
committed
533b436
Handle Double Quotes and Newline Chars in Gemma-4 Tool Call Arguments (#7477)
mamei16
authored
fbd95bd
Logits display improvements (#7486)
wiger3
authored
75bf2fe
Fix: wrong chat deleted when selection changes before confirm (#7483)
lawrence3699
authored
c7ca2f2
UI: Add sky-blue color for quoted text in light mode (#7473)
Th-Underscore
authored
ed61081
Fix version metadata not syncing on Continue (closes #7492)
oobabooga
committed
a8a0f8d
Fix row_split not working with ik_llama.cpp (closes #7489)
oobabooga
committed
9e32342
Merge remote-tracking branch 'refs/remotes/origin/dev' into dev
oobabooga
committed
160a5e1
UI: Deduplicate auto-scroll calls with queueMicrotask
oobabooga
committed
2aee3a1
Attempt at fixing chat scroll getting stuck on thinking blocks (#7485)
oobabooga
committed
4bce4a3
Update README
oobabooga
committed
4e978dd
Lint
oobabooga
committed
b06b36a
Reset bos/eos tokens to defaults before reading model metadata
oobabooga
committed
734d3c3
Merge pull request #7494 from oobabooga/dev
oobabooga
authored
9055937
UI: Add an error message when no model is loaded
oobabooga
committed
f66a01b
Merge pull request #7495 from oobabooga/dev
oobabooga
authored
eeb3288
Fix "Start reply with" crash (closes #7497)
oobabooga
committed
5992fa9
Fix tool responses with Gemma 4 template (closes #7498)
oobabooga
committed
e4168ee
UI: Fix consecutive thinking blocks rendering with Gemma 4
oobabooga
committed
dd77820
Clean up unused function
oobabooga
committed
a56e281
Merge pull request #7501 from oobabooga/dev
oobabooga
authored
7bae526
Fix bos/eos tokens being overwritten after GGUF metadata sets them (closes #7496)
oobabooga
committed
57be34d
Merge pull request #7502 from oobabooga/dev
oobabooga
authored
841aded

Commits on Apr 17, 2026

Add model download branch handling in download_model_wrapper
Th-Underscore
committed
b646090
Remove double space
Th-Underscore
committed
d6d9720

Commits on Apr 18, 2026

Commits on Apr 19, 2026

Commits on Apr 20, 2026

Add stdio MCP server support via user_data/mcp.json
oobabooga
committed
707b5df

Commits on Apr 22, 2026

Cache MCP tool discovery to avoid re-querying on each generation
oobabooga
committed
1e1d2d3
Add tool call confirmation with inline approve/reject buttons
oobabooga
committed
a253d38
Update llama.cpp
oobabooga
committed
4e065e6
llama.cpp: Pass --draft-min 48 by default for draftless speculative decoding
oobabooga
committed
9a9f3ae
Add `preserve_thinking` chat template parameter (UI and CLI)
oobabooga
committed
83347ed
Security: Fix SSRF vulnerabilities in URL fetching
oobabooga
committed
4737744
Fix Gemma 4 thinking tags not hidden after tool calls (closes #7509)
oobabooga
committed
1d871c0
Follow-up to a253d38953f3a1de298d75d0e6dd1135eae0f35f
oobabooga
committed
cb023a9
Update README
oobabooga
committed
1bb6ae4

Commits on Apr 23, 2026

Fix GPT-OSS channel tokens leaking in UI after tool calls
oobabooga
committed
7338120
Update README
oobabooga
committed
8ac5162
Update llama.cpp
oobabooga
committed
5a9d61e
Merge pull request #7512 from oobabooga/dev
oobabooga
authored
1d35a48
llama.cpp: Fix multimodal by using server's random media marker
oobabooga
committed
1d1a319
Update the custom gradio wheels
oobabooga
committed
d440d83
Pre-load MCP stdio server tools at startup
oobabooga
committed
dc798aa
Merge pull request #7514 from oobabooga/dev
oobabooga
authored
facd16c
Revert "Pre-load MCP stdio server tools at startup"
oobabooga
committed
d31662a
Merge pull request #7515 from oobabooga/dev
oobabooga
authored
8399fc4
Update transformers to 5.6
oobabooga
committed
c014efd
Update the screenshots
oobabooga
authored
7847d72
Merge branch 'main' into dev
oobabooga
committed
08b53f8

Commits on Apr 25, 2026

Fix race condition and clean up exllamav3 module
oobabooga
committed
187b07a

Commits on Apr 27, 2026

Commits on May 2, 2026

Major UI overhaul
oobabooga
committed
c6a57a4
Add Electron desktop app to portable builds
oobabooga
committed
cfda099
UI: refinements after c6a57a45727b197090d11ec4c6c28e4ce581c22d & electron user_data path
oobabooga
committed
ddec64a
Fix Stop button being ignored during tool call approval
oobabooga
committed
8e0b02a
Minor color improvement
oobabooga
committed
03b964a
Update llama.cpp
oobabooga
committed
8e7ecab
Add missing electronjs file + small fixes
oobabooga
committed
d9a83d8
Update README
oobabooga
committed
a4838ae

Commits on May 3, 2026

Improve the chat Stop button
oobabooga
committed
30177bb
Replace DuckDuckGo scraping with ddgs library
oobabooga
committed
7facf66
Update README
oobabooga
committed
717b17c
Fix extension settings not saving for user_data/extensions (closes #7525)
oobabooga
committed
8376284
Update README
oobabooga
committed
9c85092
Update README
oobabooga
committed
3f97f5c
Update README
oobabooga
committed
365848f
Update README
oobabooga
committed
25d6004
Update README
oobabooga
committed
d38124b
Cleanup modules/web_search.py
oobabooga
committed
2ca61f2
Merge pull request #7528 from oobabooga/dev
oobabooga
authored
23581cd
Electron: Fix blank window on Linux
oobabooga
committed
2db59ad
Merge pull request #7529 from oobabooga/dev
oobabooga
authored
10d32a7
Electron: keep Windows console + silence VAAPI warning
oobabooga
committed
9c08e68
Merge pull request #7531 from oobabooga/dev
oobabooga
authored
556c5b5
Electron: fix sandbox crash on stricter Linux configs
oobabooga
committed
d764aaf
Merge pull request #7532 from oobabooga/dev
oobabooga
authored
1dcde41
Electron: fix missing log colors on Windows
oobabooga
committed
31c2a82
Remove "Shutting down TextGen gracefully." log message
oobabooga
committed
233318f
Electron: persist window bounds and maximize state
oobabooga
committed
14c8301

Commits on May 4, 2026

Commits on May 5, 2026

Gemini/DeepSeek-like chat composer (closes #7534 for real this time)
oobabooga
committed
61241cf
Small light theme color change
oobabooga
committed
acacede
Update README
oobabooga
committed
59411f0
Update README
oobabooga
committed
1037bdc
Update README
oobabooga
committed
210c65a
Update README
oobabooga
committed
cd3ac27

Commits on May 6, 2026

Fix big character picture failing to load in Electron (#7540)
oobabooga
committed
4de6027
Fix truncation length reverting after model load on UI reload (closes #7540)
oobabooga
committed
71525d5
Update llama.cpp
oobabooga
committed
8536d58

Commits on May 7, 2026

Fix --listen mode in the new Electron launcher
oobabooga
committed
bf50e7e
Fix speculative decoding broken by llama.cpp arg rename (closes #7541)
oobabooga
committed
3e38f44
Add a `--no-electron` flag
oobabooga
committed
4d3f355
Electron: disable spellcheck
oobabooga
committed
bde878f
Add a scroll animation when new chat messages are sent (Gemini-like)
oobabooga
committed
5c0fab6
Don't clear chat input when no model is loaded (closes #7542)
oobabooga
committed
88d47dc
Increase bottom padding of the last chat/chat-instruct message
oobabooga
committed
206b9d2
Update llama.cpp
oobabooga
committed
4549fc6
Update README
oobabooga
committed
3ce136e
Update README
oobabooga
committed
3b9d0fe
Merge pull request #7543 from oobabooga/dev
oobabooga
authored
1f367a2